Compare commits

...
Author SHA1 Message Date
github-actions[bot] 0190fbbbf6 chore(main): release 0.4.0 (#131)
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
2026-06-30 21:09:16 +08:00
beardthelion 86341d195d test(node): guard that every repo-scoped handler is visibility-gated (#122)
* test(node): guard that every repo-scoped handler is visibility-gated

Mirrors the existing every_in_scope_mutation_has_its_gate source-string CI test on
the egress side: scrapes every repo-scoped handler (any pub/pub(crate) async fn
taking Path<(String,String..)>) and asserts it carries an authz marker, either a
read gate (authorize_repo_read / visibility_check) or a write/owner/self gate
(require_repo_owner / require_owner / did_matches / &auth.0), or is listed in
KNOWN_UNGATED with its tracking issue. Verb-agnostic on purpose: a repo-scoped read
named outside list_/get_ (git_info_refs, replicate_encrypted_blobs, withheld_paths)
is still checked, so a new ungated egress of any name fails CI instead of shipping
as the next leak.

A completeness scan over src/api/ asserts every module declaring a repo-scoped
handler is wired into the guard, so a whole new file cannot add an ungated handler
the scrape never looks at. KNOWN_UNGATED is the live gap list (certs/issues/labels/
bounties/stars per #120, plus events/webhooks/replicas/protected pending PR #113);
the staleness assert forces each entry out the moment its gate lands. A count floor
trips if the scrape silently breaks, and unit tests pin handler_names and
is_repo_scoped.

Verified by execution: removing an allowlist entry trips the missing-gate assert;
listing a gated handler trips the staleness assert; dropping a source file trips the
completeness scan; breaking the parser trips the count floor; and a non-list/get
handler (git_info_refs) is confirmed scraped and gate-checked.

* test(node): don't let a service-conditional gate satisfy the egress guard (#122)

The guard counted a handler as gated on bare substring presence of a gate
marker, so git_info_refs false-passed on visibility_check( even though that
check only runs under `if service == "git-upload-pack"` and leaves the
git-receive-pack advertisement ungated. Add gate_runs_unconditionally, which
ignores markers that sit only inside an `if service ==` block, and track
git_info_refs in KNOWN_UNGATED until #119 makes the gate unconditional. When
that lands, the staleness assert forces the entry's removal.

* fix(review): harden gate_runs_unconditionally against unclosed blocks (#122)

Code review of the egress-guard hardening:
- Clamp the span-advance index so an unclosed `if service ==` block can't
  slice past the body end and panic (latent: real source is balanced, but the
  scraper should not crash on a pathological future body). Keep span-to-EOF as
  the fail-safe direction.
- Document that only `if service ==` is detected, not `match service`, so a
  future match-arm gate isn't silently treated as full.
- Add unit cases: gate inside both service blocks (conditional), gate inside
  and outside (full), and the unclosed-block no-panic regression.
2026-06-30 20:48:03 +08:00
Kevin Codex 8b9ceec25e feat(node): iCaptcha-aware repo propagation gate with quarantine (#125)
Phase 2 of the iCaptcha work: extend the proof-of-intelligence gate from
direct writes to federated propagation, so a node that enforces iCaptcha
can defend against spam repos that originate on a node which does not.

The proof now travels with the repo and is re-verified at mirror time:

- Capture & persist: create_repo/fork_repo store the verified proof
  (raw token + claims) against the new repo. New repo_icaptcha_proofs table.

- Serve: GET /api/v1/repos/{owner}/{repo}/icaptcha-proof returns the stored
  token, gated on whole-repo "/" read like the other replication endpoints.

- Admit at mirror time: the sync worker fetches the origin's proof and
  icaptcha::admit_mirror re-verifies it offline — signature + level +
  owner-DID binding, but NOT expiry — and consumes the jti in the local
  ledger so one solved challenge admits at most one mirror per node.

- Quarantine: in enforce mode a mirror that fails the gate is kept on disk
  but hidden from clone/serve/push (authorize_repo_read + the git serve and
  receive-pack handlers) and from listings (dedup CTE + federated list), via
  a new repos.quarantined column. Shadow logs only; off admits normally.
  Fail-safe: off/shadow/no-key always admit.

Hardening from review:
- git-receive-pack also rejects quarantined mirrors (no writes while hidden).
- mirror admission keys "first-time" on DB-row absence, not on-disk presence,
  so a clone that crashed before the DB insert is still gated on retry.
- mirror jti is spent against a fixed forward retention window, not the
  proof's already-past exp, so the replay record isn't swept early.

Known limitation: an iCaptcha proof binds to the owner DID, not the repo, so
a malicious origin can get one spam repo per harvested public proof admitted
(per-node jti single-use bounds the amplification). Fully binding a proof to
a repo needs a repo-scoped iCaptcha challenge (protocol change, future work).
The operator release HTTP endpoint is deferred (no admin-auth surface yet);
DB methods are in place.

Tested: verify_token expiry-skip + signature/level/DID enforcement on the
propagation path; proof round-trip; quarantine set/release, preservation
across re-sync, exclusion from listings; mirror jti retention survives a
sweep at the proof's exp. Full workspace green against Postgres.
2026-06-30 20:46:54 +08:00
Kevin Codex 12bfb1b3c8 fix(release): use simple release-type + generic Cargo.toml updaters (#130)
release-type: rust at path . is incompatible with this virtual
workspace root. release-please's Rust strategy unconditionally appends a
CargoToml update for the root Cargo.toml, and that updater throws
"is not a package manifest (might be a cargo workspace)" whenever the
manifest has no [package] section — so every run died on the final step.

Switch to release-type: simple (touches only a version file + CHANGELOG,
never a manifest) and bump each crate version through the generic updater
via extra-files plus an inline x-release-please-version annotation.

Preserves the single vX.Y.Z tag, all action outputs, and the entire
downstream release.yml unchanged.
2026-06-30 20:01:42 +08:00
Kevin Codex 788a8686c1 fix(release): give crates explicit versions for release-please (#129)
release-please's Rust workspace strategy versions each member crate in its own
Cargo.toml and cannot rewrite an inherited `version.workspace = true`, so it
failed with 'value at path package.version is not tagged'. Give each of the five
crates an explicit version = "0.3.9" (kept in lockstep by release-please) and
drop the now-unused [workspace.package].version. Inter-crate deps are path-only,
so no version requirements need syncing.
2026-06-30 18:24:38 +08:00
beardthelion 174f25a206 fix(node): gate GET /ipfs/{cid} on per-caller path-scoped visibility (#110) (#128)
* fix(node): gate GET /ipfs/{cid} on per-caller path-scoped visibility (#110)

get_by_cid served any git object by raw SHA-256 with no auth and no
visibility check. For a public repo with a path-scoped deny rule, an
unauthenticated caller could recover a withheld blob's oid from the
served trees, derive its CID, and fetch the cleartext here, bypassing
the pack-path withholding.

Bind optional caller identity (optional_signature on a dedicated
/ipfs/{cid} sub-router; the pins route stays unsigned, tracked by #121)
and gate each iterated repo row against its OWN rules: visibility_check
at "/" denies the repo, and withheld_blob_oids skips a blob withheld
from the caller when the row carries path-scoped rules. The row is gated
directly rather than via authorize_repo_read, whose fuzzy get_repo
re-resolve could authorize a different physical row than the one read.
The per-repo withheld set is memoized on repo.id within the request.

Adds get_by_cid to the authz_guard drift table (visibility_check marker).

Scope: this closes the direct unauthenticated scan. A stale-public
mirror row still serves withheld content on both this path and the pack
path; that pre-existing leak is tracked separately in #124.

* fix(review): batch visibility-rule lookup and fail-closed walk panic in get_by_cid

Code-review follow-ups for #110:
- Replace the per-repo list_visibility_rules call with a single
  list_visibility_rules_for_repos batch query (the existing #97 pattern),
  removing the N+1 on an unauthenticated scan and the "one repo's DB error
  500s the whole request" semantics.
- Fail closed on a spawn_blocking JoinError (walk task panic) by skipping the
  repo, matching the walk-error arm, instead of 500-ing the whole scan.
- Add a regression test for the repo-level "/" deny branch (private repo,
  anon denied / owner served) and note the access gate in the module doc.

* test(node): cover the fail-closed walk-error arm in get_by_cid

Induce a withheld_blob_oids walk error (a ref pointing at a non-tree-ish
blob, same technique as visibility_pack::fails_closed_when_a_ref_cannot_be_traversed)
and assert the handler skips the whole repo: the withheld blob 404s with no
leak, and the public blob in the same repo also 404s (proving fail-closed-skip
rather than serve-on-error). Verified load-bearing: mutating the Ok(Err) arm to
treat the error as no-withholding makes it RED (secret served 200).
2026-06-30 18:16:33 +08:00
Kevin Codex 3f08d9936b Release automation (#127)
* ci(release): add Windows target and cross-platform packaging

Add x86_64-pc-windows-msvc to the release matrix. Windows ships a .zip
(7z) while unix keeps .tar.gz; a portable sha256 helper handles
sha256sum/shasum/certutil. Smoke test and packaging now resolve the
.exe suffix on Windows. Release attach uses globs to cover both formats.

* feat(release): vendor npm wrapper and auto-publish to npm

Vendor the @gitlawb/gl npm wrapper (optionalDependencies platform-package
pattern) into npm/, replacing the standalone gl-npm repo. A new npm-publish
job downloads the four unix release tarballs from the GitHub release, lays in
the binaries, stamps every package.json with the release version, and runs
npm publish --provenance. Guarded on NPM_TOKEN so it no-ops until the secret
is configured.

* feat(release): auto-bump Homebrew tap on release

Add a homebrew-bump job that regenerates Formula/gl.rb in Gitlawb/homebrew-tap
from the release tarballs' sha256 checksums (macOS+Linux, arm64+x64) and pushes
the bump. Guarded on HOMEBREW_TAP_PAT so it no-ops until the secret and tap repo
exist. Enables 'brew install gitlawb/tap/gl'.

* feat(release): Windows installer and web auto-sync on release

Add install.ps1 (Windows PowerShell installer, source of truth in node).
Add a web-sync job that opens a PR to Gitlawb/web on each release syncing
install.sh + install.ps1 + public/version.json, plus idempotent drift fixes
(remove orphaned public/bin, fix stale skill.md releases link). Guarded on
WEB_SYNC_PAT. Vercel deploys on merge.

* docs(readme): document npm, Homebrew, and Windows install methods

* ci(release): build only CLI binaries on Windows

The gitlawb-node daemon targets Linux/macOS; building it on Windows risks
a compile failure that would block the whole release (npm/brew/web jobs all
depend on release-binaries). Windows now builds only gl + git-remote-gitlawb;
the smoke-test and package loops follow the per-target binary list.

* fix(release): reconcile release-please manifest with v0.3.9 and bump action to v5

The manifest still said 0.3.8 while the only tag is a manual v0.3.9 and
Cargo.toml is 0.3.9, so release-please had no valid baseline (no v0.3.8 tag)
and never cut a release. Set the manifest to 0.3.9 so release-please diffs
v0.3.9..main and opens a proper release PR for the next version. Also pin
release-please-action to v5.0.0 to match the openclaude pipeline.

* ci(release): set explicit minimal permissions on homebrew-bump and web-sync

CodeQL flagged both jobs for lacking a permissions block, leaving GITHUB_TOKEN
at broad default scopes. Both only need contents: read (cross-repo writes use
dedicated PATs).

* fix(release): harden installers and workflow per review

- install.ps1: fail closed if the .sha256 cannot be fetched; pure ASCII (no BOM)
- npm install.js: fail (exit 1 / throw) when the platform payload or a binary is
  missing instead of reporting a successful but broken install
- workflow: top-level permissions {} + persist-credentials: false on read-only
  checkouts
- docs: PowerShell-only Windows note; text language tag on npm/README fenced block

* fix(installer): require core CLI binaries on Windows

install.ps1 only failed when zero binaries were present, so a partial archive
missing gl.exe could still report success. Require gl.exe + git-remote-gitlawb.exe
(gitlawb-node.exe stays optional, matching the Windows build) and fail early if a
required binary is absent.

* ci(release): make Windows non-blocking and stop persisting cross-repo PATs

- release-binaries: continue-on-error for the Windows leg so a failure there
  no longer fails the job or blocks npm/Homebrew (matches the best-effort intent)
- homebrew-bump/web-sync: persist-credentials: false on the PAT-backed checkouts;
  supply the token only on the git push via an x-access-token URL

* fix(release): gate web-sync on binaries and detect native arch in installer

- web-sync now needs release-binaries so the site never advertises a release
  whose artifacts failed to build
- install.ps1 reads PROCESSOR_ARCHITEW6432 before PROCESSOR_ARCHITECTURE so a
  32-bit PowerShell host on x64 Windows is not wrongly rejected

* ci(release): make npm publish idempotent across reruns

Skip any @gitlawb package version already on the registry so rerunning the
job after a partial publish no longer errors on the first existing package
and leaves the wrapper/platform packages out of sync.

* fix(release): harden web-sync and tap commit per review

- web-sync: verify the Windows ZIP + .sha256 exist on the release before
  syncing install.ps1/version.json (Windows is best-effort, so a failed leg
  must not make the site advertise a missing installer)
- stage generated files before the no-op check so brand-new untracked files
  (formula, version.json) are committed instead of skipped
- gh pr create: only swallow the 'PR already exists' case; surface auth/API errors
- npm README: scope npm/yarn/pnpm/bun install to macOS/Linux/WSL, point native
  Windows at the PowerShell installer

* ci(release): guard npm-publish before doing any work

Check NPM_TOKEN in an up-front guard step (like homebrew-bump/web-sync) and
skip checkout, release downloads, extraction, version rewrites, and publish
when it is absent, instead of doing all that work and bailing in Publish.

* fix(release): robust web PR error handling and Windows installer hardening

- web-sync: capture gh pr list output so a real API/auth failure surfaces
  instead of being misread as 'PR already exists'
- install.ps1: enforce TLS 1.2 and use -UseBasicParsing so downloads work on
  stock Windows PowerShell 5.1; remove binaries no longer shipped on upgrade so
  an in-place update can't leave a mixed-version install; handle an empty user
  PATH without a leading separator
2026-06-30 16:56:58 +08:00
Kevin CodexandClaude Opus 4.8 adc20f9eff feat(node): iCaptcha proof-of-intelligence gate on create_repo + register (#108)
* feat(node): iCaptcha proof-of-intelligence gate on create_repo + register

Add an optional gate that requires callers to present an iCaptcha proof (an
Ed25519-signed token from icaptcha.gitlawb.com, earned by solving an escalating
challenge) on the two spam-prone endpoints. Proofs are verified OFFLINE with
ed25519-dalek against the published public key, and bound to the authenticated
agent DID so they cannot be shared.

Backward-compatible by default: ICAPTCHA_MODE is `off` unless set, so deploying
changes nothing. `shadow` verifies and logs would-be rejections but always
allows; `enforce` rejects requests lacking a valid, strong-enough proof. If the
public key cannot be loaded at startup the gate stays inert (fail safe) so an
iCaptcha outage can never break repo creation or registration.

- New src/icaptcha.rs: env-driven verifier (OnceLock), /v1/pubkey fetch at
  startup or ICAPTCHA_PUBKEY override, proof parse + signature + expiry + level
  + DID-binding checks; mode-aware decide() split out for testing.
- Wired into create_repo and register (X-ICaptcha-Proof header). git push
  (receive-pack) is intentionally NOT gated.
- 11 unit tests incl. verification of a real captured proof (JS-signed,
  Rust-verified), tamper/expiry/wrong-DID/level/missing-header rejections, and
  the off/shadow/enforce/fail-safe mode matrix.

Author: Kevin Codex <[email protected]>

* fix(node): address review on icaptcha gate (validate config, bound key fetch)

Addresses CodeRabbit and @jatmn review feedback on #108:

- parse_mode now returns Option and rejects unrecognized ICAPTCHA_MODE values;
  init() warns and falls back to off instead of silently disabling the gate on
  a typo like "enforced". [P2]
- ICAPTCHA_REQUIRED_LEVEL: warn on a non-empty unparseable value instead of
  silently lowering the threshold to the default. [P2]
- fetch_key uses a reqwest client with a 5s timeout so a hung /v1/pubkey can
  never block node startup; still fails inert on timeout/error. [P2]
- .env.example: reorder ICAPTCHA_* keys (MODE, PUBKEY, REQUIRED_LEVEL, URL) to
  satisfy dotenv-linter. [P3]
- add parse_mode unit test (rejects junk, accepts documented values).

cargo test -p gitlawb-node icaptcha: 12 passed. clippy + fmt clean.

Author: Kevin Codex <[email protected]>

* fix(node): consume iCaptcha proof jti so a proof can't be replayed

Addresses @jatmn P2 on #108: the verifier checked exp/level/sub but never
consumed the proof's jti, so one solved challenge was a reusable bearer proof
for that DID until expiry, covering every gated create_repo/register in the
window. Now, in enforce mode, each verified proof's jti is recorded once and
replays are rejected.

- db: migration v7 adds icaptcha_consumed_proofs(jti PK, expires_at); new
  consume_proof_jti (INSERT ON CONFLICT DO NOTHING -> bool) and
  sweep_expired_proofs. DB-backed so it holds across restarts and instances.
- icaptcha: ProofClaims carries jti; verify returns the claims; decide() stays
  pure and returns Allow | Reject | Consume{jti,exp}; check() is now async and
  spends the jti via the DB, rejecting replays (enforce only; shadow observes
  without consuming).
- handlers: create_repo/register await check(&state.db, ...).
- main: periodic sweep of expired proof rows alongside rate-limit cleanup.
- tests: 12 pass incl. decide() yields Consume with the proof's jti.

Author: Kevin Codex <[email protected]>

* fix(node): consume iCaptcha proof only after request passes validation

Addresses CodeRabbit (Major) on #108: check() spent the jti as soon as the proof
verified, but create_repo/register ran it before name/DID/existence checks, so a
rejected request (bad name, already-exists, wrong DID) permanently burned a
valid proof without the action succeeding.

Split verification from consumption:
- icaptcha: replace check() with verify_request() -> ProofGuard (verifies and
  rejects invalid/missing proofs early; off/shadow/inert yield a no-op guard)
  plus ProofGuard::consume(db) which spends the jti right before the write and
  rejects replays. Guard is #[must_use].
- create_repo/register: verify_request() up front, then consume() immediately
  before the first write, after cheap validation passes.

Replay protection unchanged (consume-before-write). 12 tests pass; clippy clean.

Author: Kevin Codex <[email protected]>

* fix(node): give iCaptcha rejections a dedicated error code

Addresses CodeRabbit on #108: iCaptcha failure paths returned
AppError::Unauthorized, which maps to the "not_an_agent" API code and conflates
a valid-but-unproven agent with a failed signature. Add a dedicated
AppError::IcaptchaProofRequired variant mapped to the "icaptcha_proof_required"
code (HTTP 401), and use it for both the missing/invalid-proof and replay paths
so clients can distinguish "solve a challenge" from "your signature is invalid".

12 tests pass; clippy + fmt clean.

Author: Kevin Codex <[email protected]>

* fix(node): gate fork_repo with iCaptcha and cover the proof ledger

Address remaining PR #108 review comments:

- fork_repo was the third repo-creation entrypoint and bypassed the
  iCaptcha gate. Verify the proof up front (fail fast on missing/invalid)
  and consume it only just before the git clone, so cheap rejections
  (bad name, conflict, withheld subtree) never burn a valid proof —
  matching create_repo's verify/consume split.

- Add #[sqlx::test] coverage for the single-use ledger that the gate's
  anti-replay value depends on: consume_proof_jti (first-use vs replay vs
  distinct jti) and sweep_expired_proofs (only expired rows removed,
  unexpired spent proofs still block replays).

- Return 403 Forbidden (was 401) for IcaptchaProofRequired: the caller is
  authenticated but forbidden without a valid proof. The distinct
  icaptcha_proof_required code is unchanged.

- Recommend pinning ICAPTCHA_PUBKEY in production in .env.example to
  avoid the startup fetch dependency.

The two shadow-mode review comments were already resolved by the earlier
verify/consume refactor (decide/verify are IO-free; check_proof_jti gone).

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <[email protected]>
2026-06-30 10:58:08 +08:00
beardthelion 828dd279a2 fix(node): gate repo-listing and stats surfaces on visibility (#97, #99, #101, #104) (#111)
* fix(node): NFC-normalize visibility matcher to close NFC/NFD under-withholding (#101)

A deny rule authored in NFC byte-compared unequal to a path committed in NFD
(or vice versa), so blob_paths/withheld_blob_oids failed to classify the blob
and it shipped in cleartext on the replication/pack path. Normalize both the
glob prefix and the candidate path to NFC at the single matcher seam
(glob_matches + specificity); NFC not NFKC so compatibility forms are not
folded into distinct paths. Live serve is unaffected (git resolves byte-exact;
the gate only becomes more restrictive). Adds a matcher unit test and a
replication-path integration test with the variant byte inside the
rule-covered directory name.

* fix(node): fail closed on the full-scan pin path to stop dangling-blob leak (#99)

The withheld set is computed over reachable objects (rev-list --all + HEAD),
but the full-scan pin fallback (cat-file --batch-all-objects) includes dangling
objects the reachable walk never classified. Subtracting the reachable-only
withheld set therefore could not catch a private blob orphaned by a
force-push/amend, which pinned to IPFS/Pinata in cleartext.

Invert the default for blobs on the full-scan path: a candidate replicates only
if it is a non-blob (commits/trees are structural) or a reachable,
visibility-allowed blob; anything else (a dangling blob, or a withheld one) is
dropped. The delta path keeps the cheaper reachable-only subtraction since delta
candidates are always reachable, so no extra walk on the hot path.

Adds list_all_objects_with_type/all_blob_oids (typed enumeration),
replicable_blob_set + replicable_objects_fail_closed, and threads a full_scan
flag through resolve_candidates_for_push. Pin functions now take a pre-filtered
object list. Regression tests: dangling blob excluded end-to-end, fail-closed
unit semantics, and the all-blob universe includes dangling + excludes
non-blobs.

* fix(node): gate repo-listing surfaces on visibility to stop private-repo enumeration (#97)

list_repos, list_federated_repos, and the GraphQL repos query returned every
repo row with no visibility check, so an unauthenticated caller could enumerate
private repos' name, owner, description, and clone URL — and X-Total-Count
leaked the private-repo count even when rows were filtered.

Each surface now keeps a row only when visibility_check at "/" allows the
caller, the same decision the per-repo content endpoints make — not a bare
is_public filter, which would diverge for is_public=false-with-root-reader and
is_public=true-with-root-deny. The "/" decision depends on owner short/full-DID
matching and JSON reader-DID membership, so it is applied in Rust over an
over-fetched set rather than pushed into SQL; the paged path derives
X-Total-Count from the filtered set (dropping the SQL window count) and
paginates after filtering, so neither the page nor the count leaks. Adds a
batch root-rule query (list_visibility_rules_for_repos) to avoid N+1, and
converts list_all_repos_paged into list_all_repos_deduped (mirror dedup
preserved, pagination moved to Rust).

Tests: anonymous hides private repo + count, owner sees own private repo,
authorized root-reader sees an is_public=false repo (proving the visibility_check
path, not is_public), and federated listing hides private from anonymous.

* fix(review): share the listing visibility seam, extract fail-closed pin helper, add listing tests

Code-review follow-up on the VIS-subtree fixes (no behavior change to the three
shipped fixes; all confirmed sound by review):

- DRY the '/' listing gate into one shared visibility::listable_at_root used by
  list_repos, list_federated_repos, and the GraphQL repos query, so the three
  surfaces enforce one rule instead of three drifting copies
  (enforce-shared-rule-on-every-reader-path).
- Extract the full-scan fail-closed pin block into fail_closed_full_scan_objects,
  mirroring replication_withheld_set's degraded-path shape; drop the duplicated
  comment and the per-site no_rules sentinel (use &[]).
- Add the three listing tests the review flagged as missing: anonymous GraphQL
  repos hides a private repo, the paged X-Total-Count path excludes the private
  count (the KTD2 exploit shape), and an is_public=true repo under a root deny is
  not listed (proves the gate is visibility_check, not a bare is_public filter).

* fix(review): route the last inline root-visibility check through listable_at_root

Round-2 review follow-up. replication_withheld_set still made the announce
decision with an inline visibility_check(.., "/") == Allow — the fourth copy of
the listing gate the seam refactor was meant to collapse. Route it through
crate::visibility::listable_at_root (caller is always None, the anonymous
replication audience), so every reader of the '/' decision now shares one seam.
Also drop an em dash from the fail_closed_full_scan_objects doc comment.

* fix(node): gate the /api/v1/stats repos count through visibility to close the count oracle (#104)

The stats endpoint returned count_repos_deduped(): an unfiltered, unauthenticated
count of every logical repo including is_public=false and mode-A hide-existence
repos. A polling observer could detect that a hidden repo was created (N to N+1),
the exact existence signal mode-A suppresses everywhere else.

Count only the repos an anonymous caller could list: over-fetch the deduped set,
batch-load visibility rules, and keep rows that pass listable_at_root(.., None),
mirroring the listing seam in api/repos.rs. The caller is always None (meta_routes
carries no auth layer). Fail closed: any DB error collapses the count to 0, since
an under-count cannot leak existence.

count_repos_deduped is retained as the tested reference for the DEDUP_CTE dedup
count (its tests pin the CASE the live list paths share); it has no production
caller after this change and is allowed dead outside tests.

Tests: bare-private (no-rule branch), hide-existence mode-A with both repos
is_public=true (the Some(rule) branch), public-under-root-deny, list-parity,
sibling-fields-preserved, and empty-DB.

* test(review): add listing visibility regression guards (#97)

Three gaps the review flagged as uncovered:
- owner-filtered listing (?owner=) still hides a private repo and its count
  from anonymous (a distinct SQL bind branch from the unfiltered path).
- offset past the visible set returns an empty page while X-Total-Count stays
  the full visible total (guards against deriving the total from the cut page).
- a logical repo present as both a canonical row (with a root-deny rule) and a
  gossip mirror row stays hidden: pins that the DEDUP_CTE picks the canonical
  survivor so the batch rule lookup finds the deny. If dedup ever picked the
  mirror (slash-form id, no rule) the gate would fall back to is_public and leak.

* fix(review): address CodeRabbit feedback on #111

- visibility_pack NFC/NFD test: accept a SHA-1 or SHA-256 object id
  (40 | 64) instead of hardcoding 40, matching the fixture guard later in
  the file, so the test is not SHA-1-only.
- federated listing test: also assert the response `count` reflects only
  the visible repos, guarding against a regression to the pre-filter total.
- add an authenticated-caller GraphQL test: an owner sees their own private
  repo via `repos`, covering the resolver's caller-DID path that the
  anonymous-only test did not exercise.
- stats: count via the no-stars `list_all_repos_deduped()` (same DEDUP_CTE)
  instead of the stars-joined variant, since the count never needs stars.

* fix(review): match owner filter to did_matches for bare-owner mirror rows (#111)

Addresses jatmn's P2. The shared listing CTE filtered owners with
`owner_did = $1 OR owner_did LIKE '%:' || $1`, so a full `did:key:z...`
query did not match a mirror-only row stored with the bare owner `z...`.
#97 routed the no-limit owner-filtered path (used by `gl repo list`)
through this SQL predicate, regressing the full-DID-against-bare-mirror
direction that `crate::api::did_matches` handled in the old Rust path.

Normalize the bound owner to its did:key short form and compare it against
the row's owner key using the same CASE the dedup grouping already applies,
so full and short forms match symmetrically and a non-key DID still matches
only exactly. The no-arg `list_all_repos_deduped()` binds NULL and is
unaffected.

Regression test: a bare-owner mirror-only row is returned when queried by
both the full `did🔑` form and the bare short form. Confirmed RED
(test fails) before the fix.
2026-06-30 08:30:32 +08:00
beardthelion 6ae316cc88 fix(node): gate fork_repo on per-caller path-scoped visibility (#98) (#109)
* fix(node): gate fork_repo on per-caller path-scoped visibility (#98)

fork_repo authorized the caller only at the repo root "/" and then
git clone --mirror'd the full source, discarding the visibility rules.
A non-owner with root read but denied a path-scoped subtree could fork
the full mirror and obtain blobs the filtered git_upload_pack path
withholds.

Refuse the fork when the caller has any withheld glob (per-caller
withheld_globs non-empty), the same decision the read path makes;
owners and authorized subtree readers are unaffected. Fail closed with
a not-found response, before repo_store.acquire, so no withheld object
is materialized. Add unit coverage for the policy and a direct-mount
integration test pinning the gate wiring.

* fix(review): note fork-gate glob-grammar dependency, import withheld_globs

Address code-review findings on the #98 fork gate:
- Document that fork_withheld_blocks's prefix-probe equivalence with the
  serve path depends on validate_path_glob keeping "/" the only whole-repo
  scope, mirroring has_path_scoped_rule's caveat (latent-coupling guard).
- Import withheld_globs and call it unqualified, consistent with the
  sibling visibility_check import in the same module.
2026-06-28 09:01:13 +08:00
beardthelion 6809201daa fix(node): anchor the real old_sha and issue a per-ref certificate (#72)
* fix(node): anchor the real old_sha and issue a per-ref certificate (#26)

The push handler discarded the real old->new transition before anchoring
and only certified the first ref of a multi-ref push, so the permanent
Arweave record was a weaker, incomplete copy of facts the node had
already parsed.

- Thread the parsed old_sha through the clone moved into the anchor task
  instead of hardcoding a 64-zero placeholder. This also fixes the SHA
  width mismatch: object ids are 40-hex SHA-1, not 64-hex.
- Issue a signed RefCertificate for every ref the push advanced, each
  carrying that ref's real old/new, rather than one cert for the first
  ref (which, on an empty push, signed a timestamp string as new_sha).

Lock the anchor-body contract with a test asserting the upload carries
the real old_sha and new_sha, never a placeholder.

Scope is the integrity subset only. Carrying the signed certificate on
Turbo, capturing the pusher signature, chaining certificates, and
gl-side verification remain open under #26.

Refs #26

* fix(node): thread real old_sha through every ref-update fan-out path

PR #72 made the Arweave/DB anchor path per-ref with the real old_sha but
left three sibling fan-out paths in git_receive_pack flattening multi-ref
pushes and/or emitting a placeholder old_sha. Bring them in line:

- gossip publish_ref_update now forwards the real old_sha (was ""), matching
  the RefUpdateEvent contract that documents all-zeros for a created ref
- HTTP /sync/notify sends one signed request per ref with its real old_sha;
  the receiver is single-ref, so a multi-ref push no longer drops refs 2..N
- GraphQL RefUpdateBroadcast emits one broadcast per ref with the real
  old_sha instead of only the first ref + a zero placeholder

No receiver, struct, or wire-schema change; the all-zeros-for-created-ref
convention is preserved on the wire. Addresses jatmn's CHANGES_REQUESTED and
the two CodeRabbit findings.

* test(node): regression-guard the per-ref sync-notify fan-out

Extract the HTTP /sync/notify send into notify_peer_of_ref (one ref) and
notify_peer_of_refs (one peer, looping refs), collapsing the deeply-nested
inline block in git_receive_pack and giving the fan-out a unit-testable seam.
The ref_name is now also logged on every warn arm.

Add mockito regression tests: a two-ref push must emit one request per ref
each carrying its real old_sha (guards against re-flattening to .first() with
a zero placeholder, the bug PR #72 fixed), and a created ref must forward the
all-zeros hash verbatim. Behavior is unchanged; this only adds the seam, the
tests, and ref_name in the logs.

* style(node): match sibling test conventions in repos tests

Prefix the two new sync-notify tests with test_ and use the _mock binding
convention (matching arweave.rs/pinata.rs). No behavior change.

* style(node): fix rustfmt import order and dedupe the zero-sha literal

- Reorder the repos test-module imports so `use crate::...` precedes the
  external `gitlawb_core` import (unblocks cargo fmt --check / CI).
- Webhook `created` check reuses the existing ZERO_SHA constant instead of
  repeating the 40-hex literal.
- Arweave test fixtures use the canonical 40-hex zero old_sha, matching the
  SHA-1 width the per-ref anchoring contract now carries.

* perf(node): run peer notification after local broadcast and anchoring

The post-push background task awaited the HTTP peer-notify fan-out before
the GraphQL subscription broadcast and Arweave anchoring. With per-ref
notification a slow or unreachable peer could delay both local subscribers
and permanent anchoring by the sum of the peer round-trips. Move the
peer-notify loop to run last; peers already learned of the push via the
gossip publish earlier in the task, so this only deprioritizes the
best-effort HTTP fallback.

* style(node): correct notify-ordering comment and reuse ZERO_SHA in test

Review follow-ups on the two prior commits:
- The peer-notify ordering comment claimed peers 'already learned via the
  gossip publish', but that publish is gated on a present p2p handle; with
  no handle (the very case the HTTP fallback exists for) it does not run.
  State the actual rationale: peer-notify is the lowest-priority best-effort
  step, ordered last so it cannot delay local broadcast or anchoring.
- The all-zeros notify test still inlined the 40-hex literal; point it at the
  ZERO_SHA constant like the production path.
2026-06-27 23:16:23 +08:00
beardthelion 3e1e9045e4 fix(node): close under-withholding via full ref scope and full-history classification (#42) (#84)
* fix(node): close under-withholding via full ref scope and reachable-only pin set

blob_paths walked only refs/heads/* and refs/tags/* and skipped silently on a
failed git ls-tree, so a blob reachable only through another namespace, or a ref
that failed to traverse, could fall out of the withheld set and ship in cleartext.
Walk every ref and fail closed on traversal error.

The pin enumerators (ipfs_pin, pinata) used git cat-file --batch-all-objects,
which includes unreachable/dangling objects that have no path and cannot be
classified for withholding. Switch them to git rev-list --objects --all so the
pin set matches the reachable graph blob_paths evaluates.

* fix(node): include HEAD in the withheld-blob walk to close pin/serve asymmetry

blob_paths classified objects from for-each-ref only, but the pin set
(git rev-list --objects --all) and git upload-pack both also expose
objects reachable from HEAD. A detached HEAD (or any HEAD pointing
outside the ref-covered set) therefore put a blob into the pin/serve
path that the withholding walk never classified — a latent cleartext
leak of a withheld blob. Walk HEAD too (guarded for unborn HEAD on an
empty repo) so the withheld set is a superset of every exposure path.

Adds a regression test: a blob reachable only via a detached HEAD must
still be withheld (fails without this change).

* fix(node): classify blobs across full history, not just ref tips

blob_paths walked only ref-tip trees (git ls-tree -r <ref>), so a blob
that existed in an older commit and was deleted or rotated at the tip
was never classified, while the pin set (git rev-list --objects --all)
and upload-pack still expose it from history. That under-withheld the
blob into the IPFS/Pinata replication paths in cleartext.

Enumerate all reachable commits (git rev-list --all + HEAD for the
detached case) and walk each commit tree, deduping (oid, path) pairs.
ls-tree per commit is kept over rev-list --objects so every path a blob
lives at is still represented for the per-path visibility check. A ref
that resolves to a non-commit now fails closed explicitly, since
rev-list --all silently skips it where the old per-ref ls-tree errored.

Regression test: a secret blob deleted at the tip but reachable in
history is withheld.

* fix(node): share one history walk and fully peel ref types in withheld classifier

Addresses two review findings on the under-withholding classifier:

- withheld_blob_recipients walked blob_paths twice (once via
  withheld_blob_oids, once directly). Extract withheld_from_pairs so a
  single history walk feeds both the withheld set and the recipient map.

- The ref-type fail-closed guard used for-each-ref %(*objecttype), which
  peels only one tag level and so misclassified a nested annotated tag of
  a commit as a non-commit, refusing the whole walk (over-withhold). List
  refnames and peel each fully with <ref>^{} through one cat-file
  --batch-check; reap the child even if the stdin write fails. Fail closed
  on a short read or any non-commit/missing target.

Adds tests: annotated + nested-tag of a commit does not fail closed,
annotated tag of a blob still fails closed, and a blob at both an allowed
and a denied path is not withheld.

* fix(node): drain cat-file stdout concurrently to close the peel deadlock

The ref-type guard peeled every <ref>^{} through git cat-file --batch-check,
writing all queries to stdin before draining stdout. On a repo with enough
output the child's stdout pipe fills, it stops reading stdin, and write_all
blocks forever — hanging every serve/pin that classifies the repo. Batching
the queries does NOT bound this: cat-file echoes the full query on a
'<query> missing' line, so output scales with refname length, and a few
hundred long-named dangling refs already exceed the pipe buffer.

Run the stdin write on a separate thread so the main thread drains stdout
concurrently via wait_with_output, which reaps the child unconditionally
before any error is surfaced; the writer is joined after the drain so it
cannot deadlock, and a vanished stdin pipe becomes a broken-pipe write error.
One pass, no batch bound, deadlock-free regardless of ref count or refname
length.

Extracted as assert_all_refs_are_commits with corrected doc attribution.
Detect the missing-object line by its trailing status word. Adds a
timeout-bounded regression test with 500 long-named dangling refs (>64 KiB of
cat-file output) that fails rather than hangs if the deadlock returns.

* fix(node): parse ls-tree -z raw paths and fail closed on non-UTF-8 to stop under-withholding

Plain `git ls-tree -r` C-quotes any path with non-ASCII or special bytes
(café.txt becomes "secret/caf\303\251.txt"), so the stored path no longer
matched a visibility rule like /secret/** and the blob was served/pinned in
cleartext. Switch blob_paths to `git ls-tree -rz` and split the NUL-delimited
records, which emit paths raw; the TAB field separator survives -z so the
per-record parse is otherwise unchanged.

Parse strictly with str::from_utf8 and fail closed on invalid UTF-8 rather than
from_utf8_lossy: a lossy decode would rewrite an invalid byte in a denied
directory name to U+FFFD, and the mangled path would again miss its deny rule
(the same under-withholding class one layer down).

Tests: a non-ASCII path stays withheld (and the public blob still replicates),
and a non-UTF-8 path fails closed.
2026-06-27 23:15:38 +08:00
beardthelion 1960f6b28f ci(triage): label fork PRs and stop auto-closing PRs that only lack an issue link (#107)
* ci(triage): run PR triage on pull_request_target so fork PRs can be labeled

Fork PRs run the triage job with a read-only GITHUB_TOKEN under the
pull_request event regardless of the workflow's declared write
permissions, so addLabels/createComment fail with 403 'Resource not
accessible by integration' for every external contributor (e.g. #103).

Switch the trigger to pull_request_target, which runs in the base-repo
context with the declared write token. The job never checks out or runs
PR head code (it only reads context.payload and the pulls/files API and
manages labels), so it carries none of the untrusted-code-execution risk
that makes pull_request_target dangerous.

* ci(stale): drop needs-issue from auto-close trigger, keep it advisory

needs-issue flags PRs that link no issue, but linking one is a SHOULD in
CONTRIBUTING (required only for protocol-level changes), and many
legitimate small PRs (docs, obvious fixes) have none. Coupling it to the
14/21-day auto-close was overreach: a substantive, well-described PR
could be closed purely for missing an issue reference.

Keep needs-issue as an advisory label and guidance nudge; leave only
needs-description (an empty/thin body, a far stronger low-effort signal)
as the auto-close trigger.

* ci(stale): align stale message with description-only close condition

The auto-close trigger narrowed to needs-description, but the stale
message still told contributors a PR could close for a missing linked
issue. Drop that clause so the warning matches what actually closes a PR.

* style(ci): replace em dashes in triage/stale contributor messages

* docs(contributing): match stale close policy to description-only trigger

Auto-close is driven by needs-description only; a missing issue link stays
advisory. Align the contributor docs with the workflow so the OR wording no
longer implies a missing issue link can close a PR.
2026-06-27 23:14:41 +08:00
beardthelion abc9ad03ac fix(node): fail closed when a recipient DID can't be resolved (#47) (#67)
* fix(node): fail closed when a recipient DID can't be resolved (#47)

encrypt_and_pin resolved recipient DIDs with filter_map, silently
dropping any that failed to resolve and sealing the envelope to the
remaining subset while still recording the full intended recipient set
as covered. A reader whose DID could not be resolved at seal time (any
did:web/did:gitlawb, or a malformed did:key) was then permanently locked
out: the dedup compares the recorded full set against the desired set,
matches, and never re-seals.

Resolve all recipient DIDs up front via resolve_all_recipients; if any
fail, log and skip the blob rather than seal to a partial set. Nothing
partial is recorded, so the dedup stays honest and the blob re-seals on
a later push once resolution is available. Add a bounded per-blob warn
plus one aggregate coverage warn so a fully-migrated did:gitlawb org
(zero encrypted-pin coverage) is one greppable line, not a scrape, and
give the previously silent read_object and pin_git_object failure arms
warn logs.

* test(node): extract plan_seal so the #47 fail-closed gate is unit-tested

The fail-closed invariant lived only in resolve_all_recipients, tested in
isolation; nothing proved encrypt_and_pin acts on it (a refactor falling
through to a partial seal would pass every test). encrypt_and_pin needs a
live Postgres Db, which CI does not provide, so it can't be tested directly.

Extract the per-blob seal/skip decision into a pure plan_seal() -> SealPlan
(no DB, no IO), behaviour-preserving, and unit-test its branches: seals only
when all recipients resolve, fails closed on any unresolvable DID, skips the
empty set, skips an unchanged tag, re-seals a changed set.

* style(node): rustfmt encrypted_pin (Seal variant + assert_eq wrap)
2026-06-26 10:33:13 +08:00
beardthelion 338ff83584 fix(node): skip withheld-walk when no path-scoped rule can withhold (#60)
* feat(node): add has_path_scoped_rule predicate for withheld-walk short-circuit

Shared predicate (rules.iter().any(path_glob != "/")) both serve and
replication paths will use to skip the per-blob withheld walk when no rule
can withhold. Doc records the "/"-gate precondition and the validate_path_glob
dependency. Adds unit tests plus a gate-aware safety invariant test.

* fix(node): skip withheld-walk when no path-scoped rule can withhold (#50)

The serve path (git_upload_pack) ran the full per-ref withheld_blob_oids
walk on every clone/fetch before falling back to a plain upload_pack, even
for repos with no path-scoped rules where the walk can never withhold
anything. Cost grew with #42 (per-ref git ls-tree -r).

Guard both call sites on has_path_scoped_rule: the serve path skips the
spawn_blocking walk entirely and serves upload_pack directly; the
replication path generalizes its is_empty() short-circuit to also cover
root-only rule sets. Safe because the whole-repo "/" gate runs before both
sites, so a denying root rule 404s/withholds upstream and never reaches the
predicate. No change to served-pack contents on the success path.

* fix(node): gate encrypt-then-pin walk on has_path_scoped_rule (#50)

The post-push encrypt-then-pin block in git_receive_pack gated the
withheld_blob_recipients call on `!r.is_empty()`, which is true for a
root-only `/` rule set. withheld_blob_recipients then runs the same
per-ref `git ls-tree` walk (withheld_blob_oids) the short-circuit is
meant to avoid, and returns an empty recipient map anyway. Gate it on
has_path_scoped_rule to match the serve and replication sites; the skip
is behavior-preserving (empty map -> encrypt_and_pin is a no-op).

Also align the two remaining call sites to the visibility_pack import
alias for consistency.

* test(node): cover the serve decision across both has_path_scoped_rule branches

Drives the git_upload_pack serve decision over a real bare repo for the
INV-2 caller (a reader allowed at whole-repo "/" but denied a path-scoped
subtree), exercising both branches the predicate selects: root-only rules
skip the walk and serve the full pack (asserting the walk would withhold
nothing, so the skip is sound), and a /secret/** rule runs the walk and
excludes the secret blob from the served set while keeping the public one.
2026-06-26 10:32:43 +08:00
beardthelion 636d3abe17 chore(deps): bump tar to 0.4.46 and rand to 0.9.3 to clear advisories (#105)
tar 0.4.45 had a PAX header desynchronization issue (GHSA-3pv8-6f4r-ffg2,
fixed in 0.4.46); rand 0.9.2 was unsound with a custom logger using
rand::rng() (GHSA-cq8v-f236-94qc, fixed in 0.9.3). Both resolve as
lockfile-only bumps with no manifest change. Build and cargo audit pass.

lru 0.12.5 (GHSA-rhfx-m35p-ff5j) stays pinned by aws-sdk-s3 (lru ^0.12.2)
and the hickory advisories remain accepted risk per .cargo/audit.toml (#76).
2026-06-26 10:32:09 +08:00
beardthelion 2f6611a30f fix(core): zeroize the derived X25519 secret (#65) (#91)
x25519_secret_from_seed derived the X25519 private scalar from the Ed25519
seed and returned it as a bare [u8; 32]. Because [u8; 32] is Copy with no
Drop (and sha2's digest output likewise has no zeroize), the scalar, the
SHA-512 digest, and the returned temporary were released without being
scrubbed, leaving secret-derived material in freed memory. This is the same
Copy-no-Drop class fixed for the raw seed in #41, on the derived secret.

Return Zeroizing<[u8; 32]>, build the scalar directly into a zeroizing
buffer so no bare secret local persists, and explicitly wipe the SHA-512
digest before it drops. Callers deref the result into crypto_box::SecretKey,
which already scrubs its own copy. No behavior change: the derived scalar and
all decryption output are byte-identical, covered by the existing
ed25519_to_x25519_keypair_agrees and seal_open_round_trip_for_recipients tests.
2026-06-25 08:31:32 +08:00
beardthelion 3e8e333aa0 fix(node): dedupe mirror and canonical repo rows on list surfaces (#6) (#73)
* fix(node): dedupe mirror and canonical repo rows on list surfaces (#6)

The non-paged GET /api/v1/repos legacy path and list_federated_repos
returned both the short-owner peer mirror row and the canonical did:key
row for the same logical repo, so profiles rendered the repo twice. Only
the paged path collapsed them, in SQL.

Add a dedupe_canonical_repos helper that groups by (normalized owner,
name), keeps the canonical non-mirror row (tie broken by earliest
created_at), and carries the group's latest updated_at onto the
survivor, matching the paged SQL dedup. Apply it at both non-paged
surfaces and cover it with unit tests.

* fix(node): dedupe GraphQL + stats repo surfaces and key mirror detection on the repo id

Addresses jatmn's review on #73 (dedup not applied on every reader path; mirror
detection keyed on a user-settable string).

- GraphQL repos and /api/v1/stats now collapse mirror+canonical rows, via new
  Db::list_all_repos_deduped and count_repos_deduped that share the DISTINCT ON
  CTE with list_all_repos_paged so the dedup rule cannot drift.
- Mirror detection keys on the structural slash-form id (written only by
  upsert_mirror_repo) instead of the description == 'mirrored from peer' string,
  in both the SQL paths and dedupe_canonical_repos.
- Deterministic survivor on a full created_at tie (id ASC) in both
  implementations.
- Legacy REST list and federated keep their method-scoped did_matches owner
  filter in Rust; it does not compose with the method-blind SQL group key, so
  those paths intentionally stay on the Rust helper.
- Adds sqlx and unit tests for the new surfaces, the structural marker, and the
  tiebreak.

* test(node): cover mirror-only, empty, and count/list-consistency dedup cases

Follow-ups from code review on the dedup change:
- list_all_repos_deduped/count_repos_deduped: mirror-only group survives, empty
  table returns empty/0, and count_repos_deduped equals the deduped list length
  (guards the two independent SQL queries against grouping-key drift).
- Document list_all_repos as the raw, non-deduped enumeration path (object
  lookup only), so it is not mistaken for a listing-surface method.

* style(node): apply cargo fmt to dedup test code

* fix(node): make the repo dedup key did:key-aware so distinct DID methods don't collapse

The dedup grouping key took the last ':' segment of owner_did
(split_part / rsplit), so two repos owned by did:key:X and
did:gitlawb:X with the same name collapsed into one logical repo on
the list, paged, count, stats, and GraphQL surfaces. That is the exact
cross-method collision the codebase already guards against in
did_matches.

Replace it with a did:key-aware key that strips a did🔑 prefix only
when the remainder is a bare id (no ':'), otherwise keeps the full DID,
reproducing did_matches/key_id as an equivalence relation: did:key:X
and a bare mirror X still collapse, while distinct methods never merge.
Applied byte-identically across DEDUP_CTE (DISTINCT ON / PARTITION BY /
ORDER BY), count_repos_deduped, the empty-page count fallback, and the
in-memory dedupe_canonical_repos, so the SQL and Rust paths agree.

The backing index lived in the already-released migration v1, so v1 is
left untouched and a new migration v7 drops idx_repos_owner_short_name
and builds idx_repos_owner_key_name on the matching expression; the
CASE must stay byte-identical to the queries or Postgres stops using it.

Tests cover both the in-memory and SQL paths: distinct methods stay
separate, bare-id and did:key forms collapse, and the residual-colon
guard keeps a malformed did:key:did:gitlawb:X distinct from the bare
method DID. 216 pass.
2026-06-25 08:25:42 +08:00
beardthelion f28fa02236 fix(security): gate webhook creation through the public-host validator (#81) (#92)
Webhook creation only scheme-checked the URL, so an authenticated repo
owner could register a hook pointed at loopback, private, or metadata
endpoints and make the node POST to them on repo events. Route the URL
through is_public_http_url, the same validator the peer announce path
already uses.

Wiring it up surfaced a bypass in that shared validator: it folded
IPv4-mapped and compatible IPv6 but not 6to4 or NAT64, so [2002:7f00:1::]
(127.0.0.1) slipped through. Fold 6to4 and the NAT64 well-known prefix,
reject the NAT64 local-use prefix and the rest of 64:ff9b::/32 outright
(the embedded v4 offset varies by prefix length, so a decode would be
error-prone), and reject 0.0.0.0/8. The change only ever rejects more,
so the peer announce handler, the DB-boundary writer, and the boot-time
prune inherit the same hardening.

Closes #81.
2026-06-25 08:24:34 +08:00
beardthelion 1af4fdf485 feat(node): pin the per-push object delta instead of re-enumerating the whole repo (#90)
* feat(node): add push-delta object enumeration module

Compute the per-push object delta (git rev-list --objects <new> --not <old>)
for the pin path instead of enumerating the whole repo, with a fully-peeled
ref-type guard (cat-file -t '<sha>^{}') that fails closed to a full scan on a
non-commit tip, any git error, or the GITLAWB_PIN_FORCE_FULLSCAN kill-switch.
The whole-repo lister moves here for the fallback and the reconciliation sweep.

Pin-enumeration scope only (N16); the withheld filter is unchanged.

* feat(node): pin the per-push object delta instead of the whole repo

Wire the push-delta enumeration into the IPFS and Pinata pin paths so pin work
scales with push size, not repo size (N16).

- U2: both pin_new_objects twins take a candidate OID set instead of scanning
  the whole repo internally; repo_path is retained for reading object bytes.
- U4: extract the anonymous replication withheld gate (caller=None, no caller
  param) into a shared helper returning (announce, withheld), so the push path
  and the future reconciliation sweep cannot drift. The per-caller read-serve
  withheld call in git_upload_pack is deliberately left separate.
- U3: the post-receive handler resolves the push delta once (off-thread),
  falls back to a full scan on a non-commit tip / git error / kill-switch, and
  passes the candidate set to both pin paths. Logs delta size and fallbacks.

The reconciliation sweep (U5) is deferred: it depends on the history-deep
blob_paths fix (#84) landing on main first.

* fix(review): log degraded pin-candidate paths, extract wiring to push_delta

Code review found the push candidate computation collapsed both a failed
full-scan fallback and a panicked spawn_blocking task to a silent empty set,
diverging from the sibling withheld path that logs and fails closed.

Extract resolve_candidates_for_push into git::push_delta (it owns the
PinCandidates dispatch and full-scan fallback), logging every degraded path
(fallback taken, fallback failed, task panicked) instead of failing silently.
Pinning nothing on a hard failure is still safe (the withheld filter always
runs; the reconciliation sweep backstops durability) but is now observable.
Moving the dispatch out of the HTTP handler also makes it unit-testable.

Add tests: tag-of-tag-of-commit peels to a Delta, and resolve_candidates_for_push
delta + non-commit-tip full-scan-fallback paths.

* fix(node): fail announce closed when withheld walk fails (#90)

A failed or panicked withheld_blob_oids walk left announce=true while
withheld=None, so a push we could not vet would still gossip ref updates,
notify peers, and write a permanent Arweave anchor. Force announce false
in that arm so the gate fails closed on both axes.
2026-06-24 11:05:19 +08:00
beardthelion 3d880cd56f fix(node): reject malformed path globs with non-trailing or empty-segment wildcards (#74) (#75)
validate_path_glob checked `ends_with("/**")` without first stripping it, so any
glob that still ended in `/**` slipped past the wildcard guard even when it held a
`*` elsewhere in the prefix (e.g. `/a*b/**`, `/a/**/**`). The prefix is fed to
glob_prefix/glob_matches as a literal, so such a rule matched no real path and
silently granted nobody, the exact silent-misconfiguration this validator exists
to prevent.

Strip the one permitted trailing `/**` first, then reject any remaining `*`.

While here, close a related hole in the same validator: `glob_prefix` greedily
trims all trailing `**` and `/`, so `//**` collapsed to `/` (whole repo) and
bypassed the explicit `/**` guard, and `//secret/**` became a dead rule whose
prefix never matches a real single-slash path. Reject empty path segments so the
validator and glob_prefix agree on the canonical form.

Adds `/a*b/**`, `/a/**/**`, `/**/**`, `//**`, `//secret/**` to the rejection test.
2026-06-24 11:04:16 +08:00
0a15e763d2 feat(node): owner-only push enforcement behind GITLAWB_ENFORCE_OWNER_PUSH (#31) (#68)
* feat(node): owner-only push enforcement behind GITLAWB_ENFORCE_OWNER_PUSH (#31)

A valid did:key signature on git-receive-pack is authentication, not
authorization: any party can generate a key, derive its DID, sign, and push to
an unprotected branch. Owner checks only ran on protected branches, so a
non-owner could write to unprotected branches of an arbitrary repo.

Add an opt-in flag GITLAWB_ENFORCE_OWNER_PUSH (default false, mirroring
GITLAWB_REQUIRE_SIGNED_PEER_WRITES) that requires the authenticated pusher to be
the repo owner on every branch. Default-off keeps live nodes unaffected.

Design follows the maintainer's review on #31:
- Authorize on the canonical AuthenticatedDid injected by require_signature, via
  Extension<AuthenticatedDid> on the handler — not a re-parse of the headers.
  The legacy branch-protection check is left as-is.
- Fail closed: an absent identity or a non-owner caller is rejected. The policy
  lives in a pure fn (owner_push_rejection) so it is unit-tested without a DB.
- The gate runs before branch protection, so one rejection yields one error body.
- Factor the duplicated owner-match into auth::is_repo_owner, reused by
  receive-pack, protect.rs, and visibility.rs (each keeps its own message).
  events.rs / peers.rs are left out — they aren't owner checks.
- Name the push gate by intent (caller_authorized_to_push) so the planned
  Phase 2 UCAN git/push capability is a pure addition, not a rewrite.

Tests: owner (full + short DID) allowed, non-owner rejected, missing-DID
rejected, flag-off legacy. Docs: .env.example + docs/RUN-A-NODE.md.

Co-Authored-By: Claude Opus 4.8 <[email protected]>

* fix(node): address review on #68 — 403 status, finish owner-match consolidation

Per @beardthelion's review:

- Return 403 Forbidden (not 400) for authorization denials: the owner-push gate,
  the branch-protection denial, and the owner checks in protect.rs / visibility.rs.
  400 reads as a malformed request and some git/CI clients retry it.
- Finish the is_repo_owner consolidation: route the branch-protection check and
  the repos list filter through the helper. The branch-protection check now uses
  the verified AuthenticatedDid (auth.0) instead of re-parsing the headers.
- Drop the second identity source on the push path entirely: the post-push event
  recording also uses auth.0, so extract_did_from_auth (and the handler's
  HeaderMap param + import) are removed.
- Tests assert the Forbidden variant (so a future status regression fails loudly)
  and add direct coverage for is_repo_owner / caller_authorized_to_push.
- Docs: warn that enabling the flag blocks delegated and CI agents until the
  Phase 2 UCAN git/push capability is honored.

Co-Authored-By: Claude Opus 4.8 <[email protected]>

---------

Co-authored-by: FrozenRaspberry <[email protected]>
Co-authored-by: Claude Opus 4.8 <[email protected]>
2026-06-24 11:03:28 +08:00
beardthelion 2153b0b7a6 fix(gl): paginate gl-clone Arweave recovery and make /encrypted-blobs parsing schema-strict (#49) (#70)
* fix(gl): make /encrypted-blobs recovery parsing schema-strict

Deserialize the /encrypted-blobs 200 body into a typed EncryptedBlobsResponse
instead of poking at a serde_json::Value and defaulting blobs to [] via
unwrap_or_default(). A server schema regression now fails closed (a hard parse
error) rather than silently yielding nothing to recover, mirroring the existing
WithheldPathsResponse convention. The sole caller still does not abort the clone
but now surfaces the error instead of swallowing it into an empty result.

Part of #49.

* fix(gl): paginate Arweave manifest discovery and rank ties by block height

Discovery hard-coded a single first:100 GraphQL query with no pagination, so a
repo with more than 100 encrypted-manifest anchors (one per push) silently lost
recovery coverage for any blob whose most recent seal fell outside that window.
Follow cursor pagination (after + pageInfo) until coverage is complete, deduping
transaction ids and bounding the loop (1000 pages / 10000 ids, degenerate-cursor
guard) so a hostile or buggy gateway cannot grow the downstream per-id fetch loop
without limit. A full page returned without pageInfo, a cap hit, or a failure
after discovery began now warns; a page-1 failure (unconfigured/unreachable
gateway) stays silent as before.

Pagination concatenates pages in gateway order, which made merge_manifests'
equal-timestamp tie-break (previously insertion-order, i.e. accidentally newest
on single-page HEIGHT_DESC) gateway-dependent. Rank candidates by the composite
key (timestamp, Arweave block height, cid): timestamp primary keeps the existing
distinct-timestamp latest-wins; block height breaks equal-timestamp ties by
anchor recency (a not-yet-mined anchor counts as newest), preserving the
pre-pagination behavior; cid is a final deterministic tiebreak for same-block
anchors. Height is parsed defensively so a present-but-unreadable height (e.g. a
gateway stringifying large ints) cannot masquerade as a pending/newest anchor.

Part of #49.

* fix(review): bound edges per Arweave page and cover degenerate-cursor/dedup

Code-review follow-ups on the #49 recovery change:
- Cap edges parsed per GraphQL page at ARWEAVE_PAGE_SIZE so a hostile gateway
  returning one oversized page cannot defeat the MAX_TX_IDS budget or flood the
  downstream per-id manifest fetch loop (adversarial review).
- Collapse the two identical gateway-failure arms in the discovery loop into a
  single warn site, and hoist PAGE_SIZE to a named const tied to the query's
  first:100 (maintainability).
- Add tests: per-page edge bound, non-advancing-cursor termination (infinite-loop
  guard), and cross-page tx-id dedup fetched exactly once (testing).

Part of #49.

* fix(review): make the degenerate-cursor test assert, and pin the page-size invariant

Round-2 review follow-ups:
- The non-advancing-cursor test bound its mock to a _-prefixed handle with
  expect_at_most(3) and never asserted it; mockito defaults assert_on_drop=false,
  so the test passed even if the guard were removed. Bind it, tighten to expect(2)
  (the guard breaks on the second page, so discovery makes exactly two POSTs), and
  assert_async.
- Correct the parse_tx_page docstring: the per-page take() bounds the TxRef vec and
  fetch loop, not the response-body deserialization (an unbounded-body concern
  shared by every gateway read here).
- Hoist the discovery query to a const and add query_first_matches_page_size so the
  query first:N literal and ARWEAVE_PAGE_SIZE are kept equal by a test rather than
  a comment.

Part of #49.
2026-06-23 22:09:17 +08:00
beardthelion 96fcfdb1ca fix(node): preserve promisor mirror mode on unknown withheld-paths lookup (#48) (#69)
Peer sync decided mirror mode from the origin's withheld-paths answer via
classify_mirror, whose _ => Plain arm collapsed both Some(empty) (genuinely
public) and None (lookup 404'd / 5xx / network / parse error) into Plain. For an
existing promisor mirror that meant fetch_repo(.., Plain) unset
remote.origin.promisor + partialclonefilter and ran git fetch --refetch, so a
transient withheld-paths outage downgraded a still-withheld mirror to a full
clone and broke syncs until the endpoint recovered.

Replace classify_mirror with resolve_mirror_mode(withheld, exists, promisor):
only Some(empty) downgrades an existing mirror; None preserves the on-disk
promisor mode. A genuine public transition always returns Some(empty), never
None, so preserving on None cannot mask one; it only suppresses the
transient-error false positive.

Add a three-valued PromisorProbe read from git config --get exit codes so a
transient probe failure (Unknown) is not mistaken for a definitive NotPromisor
and cannot itself trigger the downgrade. One warn marks the preserve path,
derived from the resolved mode so it cannot drift.

Tests cover every resolve_mirror_mode arm (incl. the regression and the
indeterminate-probe case) and the probe's Promisor / NotPromisor / Unknown
outcomes against real git repos.
2026-06-23 22:07:53 +08:00
beardthelion c9f43b0105 fix(core): route seed access through the zeroizing wrapper (#41) (#64)
* fix(core): route seed access through the zeroizing wrapper (#41)

Remove Keypair::seed_bytes(), which returned a bare [u8; 32] copy of the
Ed25519 private seed. Because [u8; 32] is Copy with no Drop, each caller's
copy was released without being scrubbed, leaving secret key material in
freed memory. The zeroizing accessor to_seed() -> Zeroizing<[u8; 32]>
already existed beside it; this makes it the sole seed accessor and migrates
the two in-crate callers (envelope open path and a crypto test) to it via
deref coercion. No behavior change to envelope encryption: the derived
X25519 secret is byte-identical, covered by the existing round-trip tests.

Adds an identity test pinning to_seed() as the only seed accessor.

* test(core): pin to_seed's Zeroizing return type, drop duplicate round-trip

The to_seed_is_sole_seed_accessor test was named for a compile-time
property its body couldn't assert, and its from_seed -> DID check
duplicated from_seed_round_trip. Replace it with a type-annotated
binding that fails to compile if to_seed ever returns a bare [u8; 32]
again, plus the determinism check, which is the only behavior it
uniquely covered.

* fix(node): migrate node_seed pin caller to to_seed()

main grew a third seed_bytes() caller (repos.rs, from #40) after this
branch was cut. Route it through to_seed() too: node_seed is moved into
a long-lived tokio::spawn pin task, so the Zeroizing wrapper scrubs the
copy when that task ends. encrypt_and_pin takes &[u8; 32]; the wrapper
deref-coerces, so the call site is unchanged.
2026-06-23 22:05:02 +08:00
Kevin CodexandClaude Opus 4.8 97d37fd541 Perf/alloc and copy cleanups (#80)
* perf: trim allocations and redundant copies on hot paths

- Cargo.toml: enable thin LTO + symbol stripping for release builds
- git-remote-gitlawb: attach pack body after signing to move (not clone)
  potentially large pack bytes, halving peak push memory
- gitlawb-attest: deserialize payload by reference instead of cloning it
- core/http_sig: compare components by &str to avoid per-item String alloc
- node/tigris: drop redundant to_vec() copy of the response body
- node/rate_limit: only allocate the key String on first insert, not per request
- node/sync: resolve the peer table once per batch instead of per item

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

* perf(sync): skip peer lookup on empty sync batch

Return early from process_batch when dequeue_pending_syncs yields nothing
— the common case on the 30s idle poll — so an empty batch no longer fires
an unnecessary list_peers() DB query (CodeRabbit, PR #80).

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

* test(sync): cover per-batch peer origin-URL resolution

Extract the peer -> origin-URL lookup from process_batch into a pure
resolve_origin_url() helper and unit-test it (match + trailing-slash trim,
unknown DID, empty peer list). This is the per-batch resolution PR #80
introduced; pulling it out of the DB-bound process_batch lets it be tested
without Postgres. No behavior change. Addresses the needs-tests label.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

* fix(tigris): use a unique temp dir per archive extraction

The fixed `.{file_name}.tmp-extract` path let two concurrent extractions of
the same repo share one temp dir, so one's remove_dir_all/unpack could clobber
the other's in-progress work. Suffix the temp dir with a fresh UUID so each
extraction is isolated; this also removes the need to pre-clear a leftover dir
(a unique name can't collide with an interrupted run). The on-failure cleanup
still removes our own dir, and success consumes it via rename (CodeRabbit).

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

* fix(tigris): serialize repo publish per path to close the swap race

Unique temp dirs let extractions unpack in parallel, but the final
remove_dir_all + rename swap could still interleave for the same local_path,
causing a nondeterministic overwrite or a rename-onto-non-empty failure. Guard
just the swap with a process-wide mutex keyed by local_path so only one
extraction publishes per repo at a time; the expensive unpack stays parallel
(CodeRabbit).

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

* docs(tigris): restore decompress_repo doc comment ordering

The earlier publish_lock insertion landed between decompress_repo's doc
comment and its fn, gluing the two doc blocks onto publish_lock and leaving
decompress_repo undocumented. Reorder so each fn keeps its own doc. No
behavior change.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

* docs(tigris): note publish_lock map is never evicted

Document the known unbounded-growth limitation of the per-path LOCKS map:
bounded by the repo count so negligible for normal use, but high-volume
deployments may want LRU/weak-ref eviction. No behavior change (CodeRabbit).

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

---------

Co-authored-by: Claude Opus 4.8 (1M context) <[email protected]>
2026-06-23 15:38:18 +08:00
a8cc33a185 fix(security): reject non-public peer URLs + prune poisoned peers (#78)
* fix(security): reject non-public peer URLs + prune poisoned peers

The unauthenticated POST /api/v1/peers/announce only validated the
http(s) scheme, so anyone could register loopback/private 'peers'
(localhost:5432, localhost:22, internal admin URLs). On every push the
node fans out signed sync-notify POSTs to all peers — turning this into
an SSRF probe and burying the real peers under junk so node-origin repos
stopped replicating. Observed on node.gitlawb.com: 126/194 peer rows were
injected junk (fake DIDs z6MkScan…/z6MkAdmin…, localhost targets).

- is_public_http_url(): reject non-http(s), loopback, unspecified, private
  (RFC1918 + ULA), link-local, and localhost/.local/.internal hosts.
  Applied at announce time.
- db.prune_non_public_peers(): boot-time sweep removing already-injected
  rows, run from main() alongside prune_self_peers — self-heals poisoned
  tables on deploy.
- tests for the validator (accept public, reject loopback/private/internal,
  incl. IPv6 [::1]).

Co-Authored-By: OpenClaude <[email protected]>

* fix(security): address PR #78 review — build, redirect SSRF, validation gaps

Resolves the P1/P2 review feedback on the peer-announce SSRF fix:

- main.rs: revert the unrelated, incomplete `storage::build()` refactor back
  to the tigris init from main (the `storage` module never existed, breaking
  the build). Keeps only the prune_non_public_peers boot call.
- Redirect bypass: set redirect::Policy::none() on all peer-facing reqwest
  clients (shared http_client, sync worker, ping_peer) so a peer cannot answer
  `302 -> http://127.0.0.1/` and bypass the public-URL check. ping_peer now
  reuses the shared no-redirect client instead of bare reqwest::get.
- Move the public-URL guard into db::upsert_peer so the bootstrap announce-back
  path (main.rs) can no longer re-poison the table after the prune.
- is_public_http_url: strip trailing-dot FQDNs (`localhost.`), fold
  IPv4-mapped/compatible IPv6 (`::ffff:127.0.0.1`, `::127.0.0.1`) to IPv4 before
  range checks, and reject CGNAT (100.64.0.0/10). Added tests for each.
- prune_non_public_peers: collect bad DIDs then DELETE ... WHERE did = ANY($1)
  in one statement, so a transient error no longer abandons remaining rows.

Follow-up (separate issue): DNS-resolution rebinding guard rejecting peer
hostnames that resolve to private IPs.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

* fix(security): satisfy rustfmt + don't drop redirect guard on builder failure

- peers.rs: collapse the CGNAT/private check onto one line (cargo fmt --check).
- sync.rs: replace the unwrap_or_else(reqwest::Client::new()) fallback with
  .expect(). The default client follows redirects, so the fallback silently
  reintroduced the SSRF vector Policy::none() exists to close (CodeRabbit).

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

---------

Co-authored-by: OpenClaude <[email protected]>
Co-authored-by: Claude Opus 4.8 (1M context) <[email protected]>
2026-06-23 14:07:14 +08:00
beardthelion 2202b0097f feat(node): enforce per-route authorization across the REST and GraphQL surface (#87)
* test(node): add HTTP-API integration harness with CI Postgres

Adds a #[cfg(test)] test-support module: a migrated AppState over a real
#[sqlx::test] pool (test_state), a DB-free variant (test_state_lazy), the
assembled router (app), and signed_request_as for injecting an authenticated
DID without RFC-9421 signing. A test-callable run_migrations reuses the
production migrate() path, and sqlx gains the macros/migrate features so
#[sqlx::test] is available. CI's test job gets a postgres service behind a
pg_isready health gate so the DB-backed tests can run.

The proving test exercises the owner gate on PUT /visibility end to end:
non-owner is rejected with 400, owner succeeds. signed_request_as sets a JSON
content-type so the Json extractor does not 415 before the handler runs.

* feat(node): enforce per-route authorization on REST mutations

Closes the missing-authorization cluster on the signed REST surface, where
'authenticated' meant 'any participant' because identities are permissionless.
Adds two shared helpers in api/mod.rs: require_repo_owner (403) and did_matches,
a method-safe DID comparison that collapses the full did:key vs bare short form
without matching across methods (did:web/did:gitlawb share the base58 space).

Per-route gates:
- merge_pr is owner-only (subsumes branch protection); close_pr and close_issue
  allow the repo owner or the PR/issue author (the issue author is read from the
  git-JSON blob, with a None author falling back to owner-only).
- create_review, create_comment, create_issue_comment, and create_bounty are
  read-gated (participants need read access, not ownership; this also closes the
  private-repo bounty leak).
- the task handlers bind the acting DID to the authenticated signer.
- dispute_bounty, previously ungated, now requires the creator or claimant; the
  other bounty and replica identity checks switch from raw equality to did_matches.

A source-level guard test asserts every in-scope mutation handler still carries
its expected gate marker, so a removed gate or an unclassified new route fails
CI. DB-backed tests (via the integration harness) cover merge owner-only and the
task signer-binding; did_matches has unit coverage for the cross-method case.

* feat(node): authenticate GraphQL mutations

GraphQL mutations were a fully unauthenticated parallel write path to the task
system (N2): the /graphql route had no auth layer and the resolvers took the
acting DID as plain arguments. Apply optional_signature to the /graphql POST so
a verified DID is attached when a signature is present (queries stay open), and
thread it into request-scoped data. Each mutation now reads the verified signer
via require_signer and binds the acting DID to it (rejecting an unsigned request
or one whose claimed delegator/assignee/by_did differs from the signer), reusing
did_matches for the comparison. Read-only subscriptions (/graphql/ws) are left
open. A resolver-level test covers the unsigned, mismatched, and matching cases.

* fix(node): path-scope the get_tree visibility gate

get_tree gated on the repo root ("/") regardless of the requested subtree, so
a caller denied a withheld subtree could still enumerate its filenames and blob
SHAs (N3). Gate on the requested path instead, mirroring get_blob, and reject
traversal segments. A cross-caller test shows the rejection is path-scoped: a
non-reader is denied the withheld subtree but passes the gate on a non-withheld
path.

* fix(node): authorize task completion and read-gate PR/issue creation

Two authorization gaps surfaced by an adversarial review of the auth stack,
both necessary-but-insufficient siblings of fixes already in it.

complete_task/fail_task bound the acting DID to the signer but never checked
the caller against the task's assignee, and finish_task updated by id alone, so
any authenticated did:key could finish or fail any task in any state. Load the
task and require the caller to be its assignee (REST and GraphQL), and add an
'AND status=claimed' predicate so only a claimed task transitions. The
now-unused by_did request fields are removed.

create_pr/create_issue resolved the repo with get_repo but skipped
authorize_repo_read, so a non-reader could open a PR (firing the owner's
webhooks) or file an issue against a private repo they cannot read. Gate both
on read access like their create_review/create_comment/create_bounty siblings.

Adds DB-backed tests covering non-assignee rejection (including the empty-body
bypass), the claimed-state predicate, and private-repo PR/issue denial.

* test(node): harden the authz drift-guard against false passes

The guard asserted a marker STRING appeared in each handler body, which a bare
identifier in a comment or log line could satisfy even after the real gate was
deleted, and its body slice over-ran on any handler not declared exactly
'pub async fn'.

- Markers are now gate-shaped: a call (require_repo_owner(, did_matches(,
  authorize_repo_read() or a binding/comparison expression
  (caller != &record.owner_did, let owner_did = auth.0), never a field name.
- Full-line comments are stripped before matching, so a gate that survives only
  as a comment no longer counts as enforced.
- The body slice bounds at the next top-level fn item across pub async,
  pub(crate) async, async, pub, and bare fn forms.
- Adds rows for the now-read-gated create_pr/create_issue and for fork_repo,
  create_repo, set_profile, claim_bounty; register is excluded with a note (it
  trusts the body did, tracked as P3 D3-1).
- Adds a self-test proving a comment-only marker does not satisfy a row.

* fix(node): P3 consistency batch from the adversarial review

Five low-severity items from the stack review. The sixth (a did_matches
false-negative on non-canonical multibase encodings of the same key) is left
as-is: it is a self-inflicted owner lockout unreachable with the conforming
client, never an impersonation vector, so decoding keys in the auth path is not
warranted.

- register bound the registered DID to the request body, not the signer, so a
  signed caller could create or refresh a trust row under a victim DID. Bind it
  to auth.0 (403 on mismatch) and add a drift-guard row now that it is gated.
- Bounty submit/approve/cancel returned 400 for an identity-mismatch denial
  where dispute returned 403. Return Forbidden, matching the rest of the surface.
- The visibility/protect owner gates (require_owner and the protect.rs inline
  checks) returned 400 for a non-owner; return 403 like require_repo_owner, and
  assert the exact code in the harness.
- get_tree rejected ./.. segments but not empty interior segments, so a path
  like secret//x could reach git while the gate saw a different string. Reject
  empty interior segments too (the empty path remains the root listing).
- Tighten the get_tree path-scope test to assert an exact 200 on the
  non-withheld branch so a future upstream 4xx/5xx cannot masquerade as
  gate-pass.

Adds a register-binding regression test.

* fix(node): unify protect/visibility owner checks on did_matches

CodeRabbit flagged that protect_branch/unprotect_branch and visibility's
require_owner used a trailing-segment owner compare (split(':').next_back())
distinct from the hardened did_matches, so the two owner-match idioms could
drift. It is not exploitable (the authenticated caller is always the full did:
form while the short form is a bare suffix, so no wrong caller passes), but it
is a real inconsistency, and the trailing-segment form is a false-negative for a
bare-stored owner. Replace both with did_matches (collapses did:key full vs bare
on both sides, never across methods), and extend the drift guard to assert
require_owner itself uses did_matches, not just that it is called.
2026-06-23 13:16:35 +08:00
beardthelion ff492b4521 feat(node): signature-gated agent self-deregister (#29) (#63)
* feat(node): signature-gated agent self-deregister (#29)

Agents whose keys are rotated or compromised could never be removed from
the registry, so an orphaned DID persisted and could win ?capability=
routing. Give the DID's key holder a way to retire their own agent.

A v5 migration adds an agent lifecycle status (default 'active') plus a
deactivated_at timestamp. list_agents drops non-active agents via a pure
filter_discoverable helper, so a revoked DID no longer appears in
discovery or capability routing; get_agent stays unfiltered so a revoked
DID still resolves and surfaces its status.

DELETE /api/v1/agents/{did} lets an rfc9421-authenticated caller mark its
own agent revoked. Authorization is on the verified AuthenticatedDid via a
pure caller_matches_did gate (a DID can only retire itself), unit-tested
including the cross-DID rejection. Revocation is terminal: register_agent's
ON CONFLICT leaves status untouched, so re-registering a leaked DID cannot
resurrect it.

Authority is team-owned (the key holder); no external contract, chain, or
operator power. Lost-key retirement and any successor pointer are out of
scope.

* fix(node): preserve original deactivated_at on repeat revoke

A retried DELETE /api/v1/agents/{did} re-ran the revoke UPDATE and
overwrote deactivated_at with a fresh timestamp, losing the true
first-retired time. Use COALESCE(deactivated_at, $2) so the original
timestamp is kept; the row still matches so the idempotent 204 is
unchanged.
2026-06-23 11:41:59 +08:00
beardthelion 2d4105e479 chore(deps): bump quinn-proto 0.11.14 -> 0.11.15 (RUSTSEC-2026-0185) (#83)
quinn-proto 0.11.14 (pulled transitively via libp2p's QUIC transport) is
flagged by RUSTSEC-2026-0185. 0.11.15 is a semver-compatible patch that
clears it; cargo audit passes clean after the bump. Lockfile-only, no
manifest or code change, so it carries no ignore entry. Repo-wide:
unblocks cargo audit on main and every open branch.
2026-06-23 11:39:18 +08:00
beardthelion 803d83efdb fix(node): reap leaked git child processes (#53) (#61)
* fix(docker): reap orphan processes via tini + init, cap pids

The node runs as PID 1 in the container and never reaps reparented
orphans, so git grandchildren (e.g. pack-objects left behind when a
served clone disconnects) accumulate as zombies until fork() fails with
EAGAIN. Run under tini and set compose init: true so reparented orphans
are reaped, and add pids_limit as a blast-radius cap.

Refs #53

* fix(node): reap leaked git child processes (#53)

Serving git over smart-HTTP could leak zombie processes that, over weeks,
exhaust the PID table until fork() fails with EAGAIN and all git
operations break until restart.

The serve path (run_git_service) now runs git in its own process group and
SIGKILLs that group when the request future is dropped (client disconnect)
or returns early, so git and its pack-objects child are torn down together
instead of orphaning the grandchild to PID 1 as a zombie. A drop guard,
disarmed once the request completes cleanly, owns this teardown.

create_issue and build_filtered_pack spawned a std Child and dropped it
unwaited if the stdin write failed (and build_filtered_pack could panic on
a missing stdin handle). Both now always wait the child before surfacing
any error, so the failure path cannot leak.

Adds libc for the group-kill syscall and three regression tests covering
direct-child kill, grandchild reaping via the group, and the disarmed
no-kill path.

Refs #53

* fix(review): reap-then-disarm on serve-path errors; tini/comment nits

Address code-review findings on the #53 fix:

- run_git_service no longer early-returns on a stdin-write error. It captures
  the write result, always reaps the child, then disarms the group guard
  before surfacing the error. Previously a write/wait error path could drop
  the still-armed guard after the child was already reaped, so the group
  SIGKILL could in theory land on a reused pgid. (correctness, reliability,
  adversarial all flagged this region.)
- Drop tini's -g flag: the node manages its own children and graceful
  shutdown, so group-wide signalling would only disturb in-flight git helpers
  on shutdown.
- Correct the SAFETY comment (kill(2) is not relevant to async-signal-safety
  here) and the alive() test-helper comment (EPERM does not read as alive).

* fix(node): SIGTERM (not SIGKILL) the served-git group on disconnect

SIGKILL gave git no chance to run its signal cleanup, so an aborted
receive-pack could leave a stale .git/*.lock that blocks the next push.
SIGTERM lets git remove its lock files before exiting. Signalling the
whole process group still reaches pack-objects directly, so the grandchild
is terminated rather than orphaned, and a single immediate signal avoids
any delayed-kill pid-reuse window.

Refs #53

* docs(node): correct stale SIGKILL comments to SIGTERM in served-git teardown
2026-06-23 10:10:17 +08:00
75 changed files with 11161 additions and 804 deletions

No files matched your search

+24
View File
@@ -72,6 +72,12 @@ GITLAWB_BOOTSTRAP_DISABLE_SEEDS=false
# rolling upgrades so existing live nodes can still communicate.
GITLAWB_REQUIRE_SIGNED_PEER_WRITES=false
# Require the authenticated pusher to be the repo owner on git-receive-pack.
# A valid did:key signature is authentication, not authorization: anyone can
# sign as their own DID. When true, pushes from a non-owner DID are rejected.
# Keep false until the repo owner is ready for owner-only writes.
GITLAWB_ENFORCE_OWNER_PUSH=false
# Comma-separated libp2p multiaddrs.
# Example: /ip4/1.2.3.4/udp/7546/quic-v1/p2p/12D3KooW...
GITLAWB_P2P_BOOTSTRAP=
@@ -87,3 +93,21 @@ GITLAWB_MAX_PACK_BYTES=2147483648
# ── Sync ─────────────────────────────────────────────────────────────────
# Enable automatic background sync from known peers
GITLAWB_AUTO_SYNC=false
# ── iCaptcha proof-of-intelligence gate ───────────────────────────────────
# Optional gate on create_repo + register: require callers to present an
# iCaptcha proof (X-ICaptcha-Proof header) earned at icaptcha.gitlawb.com.
# Default off = fully inert (no behavior change).
# off - gate disabled (default)
# shadow - verify + log would-be rejections, but always allow
# enforce - reject requests without a valid, strong-enough proof
ICAPTCHA_MODE=off
# Optional base64url Ed25519 public key override; if unset it is fetched from
# ICAPTCHA_URL/v1/pubkey at startup. Recommended in production: pinning the key
# removes the startup dependency on the iCaptcha service being reachable and
# avoids silently running inert if that one-time fetch fails.
ICAPTCHA_PUBKEY=
# Minimum proof difficulty level required to pass.
ICAPTCHA_REQUIRED_LEVEL=3
# Base URL of the iCaptcha service.
ICAPTCHA_URL=https://icaptcha.gitlawb.com
+19
View File
@@ -54,6 +54,23 @@ jobs:
toolchain: [stable, beta]
# Beta is informational: an upstream beta regression should warn, not block merges.
continue-on-error: ${{ matrix.toolchain == 'beta' }}
# Postgres for the DB-backed integration tests (#[sqlx::test] provisions an
# isolated database per test off this connection). The health-check gate keeps
# `cargo test` from racing a not-yet-ready server.
services:
postgres:
image: postgres:16
env:
POSTGRES_USER: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_DB: gitlawb_test
ports:
- 5432:5432
options: >-
--health-cmd pg_isready
--health-interval 10s
--health-timeout 5s
--health-retries 5
steps:
- name: Check out repository
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
@@ -71,6 +88,8 @@ jobs:
key: ${{ matrix.toolchain }}
- name: cargo test
env:
DATABASE_URL: postgres://postgres:postgres@localhost:5432/gitlawb_test
run: cargo test --workspace
build-release:
+8 -4
View File
@@ -1,7 +1,11 @@
name: PR Triage
# Runs on pull_request_target, not pull_request: fork PRs get a read-only
# GITHUB_TOKEN under pull_request regardless of the permissions below, so the
# label/comment writes 403. This job never checks out or runs PR head code (it
# only reads context.payload and the files API), so pull_request_target is safe.
on:
pull_request:
pull_request_target:
types: [opened, edited, synchronize, reopened]
permissions:
@@ -64,8 +68,8 @@ jobs:
if (isFork && touchesWorkflows) want.add("workflow-change");
const changedRust = names.some(n => n.startsWith("crates/") && n.endsWith(".rs"));
// A test can live in a tests/ dir, a *_test.rs file, or — the common Rust
// pattern — an inline #[test] / #[cfg(test)] block added inside the source file.
// A test can live in a tests/ dir, a *_test.rs file, or (the common Rust
// pattern) an inline #[test] / #[cfg(test)] block added inside the source file.
const addsInlineTest = files.some(f =>
f.filename.endsWith(".rs") && f.patch &&
/^\+.*#\[(test\]|cfg\(test\))/m.test(f.patch));
@@ -104,7 +108,7 @@ jobs:
const reasons = {
"needs-issue": "Link the issue this addresses (`Closes #123`). For protocol changes, open an issue first.",
"needs-description": "Add a short summary of what changes and why — the template's Summary and Motivation sections.",
"needs-description": "Add a short summary of what changes and why: the template's Summary and Motivation sections.",
"needs-tests": "This changes Rust source but no tests changed. Tests are required for fixes and strongly encouraged for features.",
};
const lines = advisory.map(l => `- ${reasons[l]}`).join("\n");
+384 -15
View File
@@ -9,6 +9,9 @@ concurrency:
group: release-${{ github.ref }}
cancel-in-progress: false
# Least privilege by default; every job opts in to exactly what it needs.
permissions: {}
jobs:
release-please:
if: ${{ github.repository == 'Gitlawb/node' }}
@@ -24,7 +27,7 @@ jobs:
steps:
- name: Run release-please
id: release
uses: googleapis/release-please-action@16a9c90856f42705d54a6fda1823352bdc62cf38 # v4.3.0
uses: googleapis/release-please-action@45996ed1f6d02564a971a2fa1b5860e934307cf7 # v5.0.0
with:
token: ${{ secrets.GITHUB_TOKEN }}
@@ -41,6 +44,7 @@ jobs:
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
with:
ref: ${{ needs.release-please.outputs.tag_name }}
persist-credentials: false
- name: Set up QEMU
uses: docker/setup-qemu-action@29109295f81e9208d7d86ff1c6c12d2833863392 # v3.6.0
@@ -101,6 +105,9 @@ jobs:
needs: release-please
if: ${{ needs.release-please.outputs.release_created == 'true' }}
runs-on: ${{ matrix.os }}
# Windows is best-effort: a failure there must not fail the job or block the
# downstream npm/Homebrew jobs that only consume the unix artifacts.
continue-on-error: ${{ startsWith(matrix.target, 'x86_64-pc-windows') }}
permissions:
contents: write
strategy:
@@ -115,11 +122,14 @@ jobs:
os: macos-13
- target: aarch64-apple-darwin
os: macos-14
- target: x86_64-pc-windows-msvc
os: windows-latest
steps:
- name: Checkout release tag
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
with:
ref: ${{ needs.release-please.outputs.tag_name }}
persist-credentials: false
- name: Set up Rust toolchain
uses: dtolnay/rust-toolchain@b3b07ba8b418998c39fb20f53e8b695cdcc8de1b # stable
@@ -143,8 +153,22 @@ jobs:
with:
key: ${{ matrix.target }}
# The gitlawb-node daemon is a Linux/macOS service; on Windows we ship only
# the two CLI binaries. A Windows-only failure must not block the release.
- name: Determine binaries
shell: bash
run: |
BINS="gl git-remote-gitlawb gitlawb-node"
case "${{ matrix.target }}" in *windows*) BINS="gl git-remote-gitlawb" ;; esac
echo "BINS=$BINS" >> "$GITHUB_ENV"
- name: Build
run: cargo build --release --target ${{ matrix.target }} -p gl -p git-remote-gitlawb -p gitlawb-node
shell: bash
run: |
set -euo pipefail
args=""
for b in $BINS; do args="$args -p $b"; done
cargo build --release --target ${{ matrix.target }} $args
# Run each packaged binary's --version so a broken release artifact fails the
# build instead of shipping. Skipped for cross-compiled targets that can't run
@@ -156,9 +180,11 @@ jobs:
set -euo pipefail
VERSION="${{ needs.release-please.outputs.version }}"
BIN_DIR="target/${{ matrix.target }}/release"
for bin in gl git-remote-gitlawb gitlawb-node; do
EXE=""
case "${{ matrix.target }}" in *windows*) EXE=".exe" ;; esac
for bin in $BINS; do
echo "== $bin --version =="
out="$("$BIN_DIR/$bin" --version)"
out="$("$BIN_DIR/$bin$EXE" --version)"
echo "$out"
# Each binary prints "<name> <version>"; assert the released version is present.
grep -qF "$VERSION" <<<"$out" || {
@@ -168,26 +194,369 @@ jobs:
done
- name: Package
id: pkg
shell: bash
run: |
set -euo pipefail
NAME="gitlawb-node-${{ needs.release-please.outputs.version }}-${{ matrix.target }}"
TARGET="${{ matrix.target }}"
NAME="gitlawb-node-${{ needs.release-please.outputs.version }}-${TARGET}"
BIN_DIR="target/${TARGET}/release"
EXE=""
case "$TARGET" in *windows*) EXE=".exe" ;; esac
# Portable sha256 → "<hash> <file>" so install scripts and brew can parse it.
sha256_file() {
local f="$1"
if command -v sha256sum >/dev/null 2>&1; then
sha256sum "$f"
elif command -v shasum >/dev/null 2>&1; then
shasum -a 256 "$f"
else
local h
h=$(certutil -hashfile "$f" SHA256 | sed -n 2p | tr -d ' \r')
printf '%s %s\n' "$h" "$f"
fi
}
mkdir -p "dist/$NAME"
cp "target/${{ matrix.target }}/release/gl" "dist/$NAME/"
cp "target/${{ matrix.target }}/release/git-remote-gitlawb" "dist/$NAME/"
cp "target/${{ matrix.target }}/release/gitlawb-node" "dist/$NAME/"
for bin in $BINS; do
cp "$BIN_DIR/$bin$EXE" "dist/$NAME/"
done
cp README.md LICENSE-MIT LICENSE-APACHE "dist/$NAME/"
cd dist
tar czf "$NAME.tar.gz" "$NAME"
shasum -a 256 "$NAME.tar.gz" > "$NAME.tar.gz.sha256"
echo "asset=dist/$NAME.tar.gz" >> "$GITHUB_OUTPUT"
echo "checksum=dist/$NAME.tar.gz.sha256" >> "$GITHUB_OUTPUT"
case "$TARGET" in
*windows*)
# 7z is preinstalled on windows-latest runners.
7z a -tzip "$NAME.zip" "$NAME" >/dev/null
sha256_file "$NAME.zip" > "$NAME.zip.sha256"
;;
*)
tar czf "$NAME.tar.gz" "$NAME"
sha256_file "$NAME.tar.gz" > "$NAME.tar.gz.sha256"
;;
esac
- name: Attach to release
uses: softprops/action-gh-release@72f2c25fcb47643c292f7107632f7a47c1df5cd8 # v2.3.2
with:
tag_name: ${{ needs.release-please.outputs.tag_name }}
files: |
${{ steps.pkg.outputs.asset }}
${{ steps.pkg.outputs.checksum }}
dist/*.tar.gz
dist/*.zip
dist/*.sha256
npm-publish:
name: Publish to npm
needs: [release-please, release-binaries]
if: ${{ needs.release-please.outputs.release_created == 'true' }}
runs-on: ubuntu-latest
permissions:
contents: read
id-token: write # npm provenance
steps:
- name: Guard on secret
id: guard
env:
NPM_TOKEN: ${{ secrets.NPM_TOKEN }}
run: |
if [ -z "${NPM_TOKEN:-}" ]; then
echo "::warning::NPM_TOKEN is not set — skipping npm publish."
echo "enabled=false" >> "$GITHUB_OUTPUT"
else
echo "enabled=true" >> "$GITHUB_OUTPUT"
fi
- name: Checkout release tag
if: ${{ steps.guard.outputs.enabled == 'true' }}
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
with:
ref: ${{ needs.release-please.outputs.tag_name }}
persist-credentials: false
- name: Lay in release binaries
if: ${{ steps.guard.outputs.enabled == 'true' }}
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
VERSION: ${{ needs.release-please.outputs.version }}
TAG: ${{ needs.release-please.outputs.tag_name }}
run: |
set -euo pipefail
# npm platform package -> Rust target triple (unix only; Windows is not
# published to npm).
MAP="
gl-darwin-arm64:aarch64-apple-darwin
gl-darwin-x64:x86_64-apple-darwin
gl-linux-arm64:aarch64-unknown-linux-musl
gl-linux-x64:x86_64-unknown-linux-musl
"
mkdir -p _dl
for entry in $MAP; do
pkg="${entry%%:*}"
target="${entry#*:}"
archive="gitlawb-node-${VERSION}-${target}.tar.gz"
echo "==> $pkg <- $archive"
gh release download "$TAG" --repo "$GITHUB_REPOSITORY" \
--pattern "$archive" --dir _dl --clobber
tar -xzf "_dl/$archive" -C _dl
src="_dl/gitlawb-node-${VERSION}-${target}"
cp "$src/gl" "npm/packages/$pkg/gl"
cp "$src/git-remote-gitlawb" "npm/packages/$pkg/git-remote-gitlawb"
chmod +x "npm/packages/$pkg/gl" "npm/packages/$pkg/git-remote-gitlawb"
done
- name: Set versions
if: ${{ steps.guard.outputs.enabled == 'true' }}
env:
VERSION: ${{ needs.release-please.outputs.version }}
run: |
set -euo pipefail
node -e '
const fs = require("fs");
const v = process.env.VERSION;
for (const p of fs.readdirSync("npm/packages")) {
const f = `npm/packages/${p}/package.json`;
const pkg = JSON.parse(fs.readFileSync(f, "utf8"));
pkg.version = v;
if (pkg.optionalDependencies) {
for (const k of Object.keys(pkg.optionalDependencies)) {
pkg.optionalDependencies[k] = v;
}
}
fs.writeFileSync(f, JSON.stringify(pkg, null, 2) + "\n");
}
'
- name: Publish
if: ${{ steps.guard.outputs.enabled == 'true' }}
env:
NPM_TOKEN: ${{ secrets.NPM_TOKEN }}
VERSION: ${{ needs.release-please.outputs.version }}
run: |
set -euo pipefail
echo "//registry.npmjs.org/:_authToken=${NPM_TOKEN}" > ~/.npmrc
# Platform packages first, then the wrapper (so its optionalDependencies resolve).
# Skip versions already on the registry so a rerun after a partial publish
# is idempotent instead of erroring on the first existing package.
for pkg in gl-darwin-arm64 gl-darwin-x64 gl-linux-arm64 gl-linux-x64 gl; do
name="@gitlawb/$pkg"
if npm view "$name@$VERSION" version >/dev/null 2>&1; then
echo "==> $name@$VERSION already published, skipping"
continue
fi
echo "==> npm publish $name@$VERSION"
npm publish "npm/packages/$pkg" --provenance --access public
done
homebrew-bump:
name: Bump Homebrew tap
needs: [release-please, release-binaries]
if: ${{ needs.release-please.outputs.release_created == 'true' }}
runs-on: ubuntu-latest
permissions:
contents: read # GITHUB_TOKEN downloads release assets; the tap push uses HOMEBREW_TAP_PAT
steps:
- name: Guard on secret
id: guard
env:
HOMEBREW_TAP_PAT: ${{ secrets.HOMEBREW_TAP_PAT }}
run: |
if [ -z "${HOMEBREW_TAP_PAT:-}" ]; then
echo "::warning::HOMEBREW_TAP_PAT is not set — skipping Homebrew bump."
echo "enabled=false" >> "$GITHUB_OUTPUT"
else
echo "enabled=true" >> "$GITHUB_OUTPUT"
fi
- name: Checkout tap repo
if: ${{ steps.guard.outputs.enabled == 'true' }}
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
with:
repository: Gitlawb/homebrew-tap
token: ${{ secrets.HOMEBREW_TAP_PAT }}
path: tap
persist-credentials: false
- name: Regenerate formula
if: ${{ steps.guard.outputs.enabled == 'true' }}
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
VERSION: ${{ needs.release-please.outputs.version }}
TAG: ${{ needs.release-please.outputs.tag_name }}
run: |
set -euo pipefail
base="https://github.com/${GITHUB_REPOSITORY}/releases/download/${TAG}"
mkdir -p _sums
gh release download "$TAG" --repo "$GITHUB_REPOSITORY" \
--pattern '*.tar.gz.sha256' --dir _sums --clobber
sha() { awk '{print $1}' "_sums/gitlawb-node-${VERSION}-$1.tar.gz.sha256"; }
SHA_MAC_ARM="$(sha aarch64-apple-darwin)"
SHA_MAC_X64="$(sha x86_64-apple-darwin)"
SHA_LNX_ARM="$(sha aarch64-unknown-linux-musl)"
SHA_LNX_X64="$(sha x86_64-unknown-linux-musl)"
mkdir -p tap/Formula
cat > tap/Formula/gl.rb <<EOF
class Gl < Formula
desc "Gitlawb CLI — decentralized git for AI agents and developers"
homepage "https://gitlawb.com"
version "${VERSION}"
license "MIT OR Apache-2.0"
on_macos do
on_arm do
url "${base}/gitlawb-node-${VERSION}-aarch64-apple-darwin.tar.gz"
sha256 "${SHA_MAC_ARM}"
end
on_intel do
url "${base}/gitlawb-node-${VERSION}-x86_64-apple-darwin.tar.gz"
sha256 "${SHA_MAC_X64}"
end
end
on_linux do
on_arm do
url "${base}/gitlawb-node-${VERSION}-aarch64-unknown-linux-musl.tar.gz"
sha256 "${SHA_LNX_ARM}"
end
on_intel do
url "${base}/gitlawb-node-${VERSION}-x86_64-unknown-linux-musl.tar.gz"
sha256 "${SHA_LNX_X64}"
end
end
def install
bin.install "gl"
bin.install "git-remote-gitlawb"
end
test do
assert_match version.to_s, shell_output("#{bin}/gl --version")
end
end
EOF
- name: Commit and push
if: ${{ steps.guard.outputs.enabled == 'true' }}
working-directory: tap
env:
HOMEBREW_TAP_PAT: ${{ secrets.HOMEBREW_TAP_PAT }}
VERSION: ${{ needs.release-please.outputs.version }}
run: |
set -euo pipefail
git config user.name "github-actions[bot]"
git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
# Stage first so a brand-new (untracked) formula is detected by the no-op check.
git add Formula/gl.rb
if git diff --cached --quiet; then
echo "Formula already up to date."
exit 0
fi
git commit -m "gl ${VERSION}"
# Credentials are not persisted in .git/config; supply the token only for the push.
git push "https://x-access-token:${HOMEBREW_TAP_PAT}@github.com/Gitlawb/homebrew-tap.git" HEAD:main
web-sync:
name: Sync web (install scripts + version)
needs: [release-please, release-binaries]
if: ${{ needs.release-please.outputs.release_created == 'true' }}
runs-on: ubuntu-latest
permissions:
contents: read # GITHUB_TOKEN checks out node; the web PR uses WEB_SYNC_PAT
steps:
- name: Guard on secret
id: guard
env:
WEB_SYNC_PAT: ${{ secrets.WEB_SYNC_PAT }}
run: |
if [ -z "${WEB_SYNC_PAT:-}" ]; then
echo "::warning::WEB_SYNC_PAT is not set — skipping web sync."
echo "enabled=false" >> "$GITHUB_OUTPUT"
else
echo "enabled=true" >> "$GITHUB_OUTPUT"
fi
- name: Checkout node (release tag)
if: ${{ steps.guard.outputs.enabled == 'true' }}
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
with:
ref: ${{ needs.release-please.outputs.tag_name }}
path: node
persist-credentials: false
- name: Checkout web
if: ${{ steps.guard.outputs.enabled == 'true' }}
uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
with:
repository: Gitlawb/web
token: ${{ secrets.WEB_SYNC_PAT }}
path: web
persist-credentials: false
# web-sync publishes install.ps1 + version.json, so the Windows ZIP must exist.
# Windows is best-effort for npm/Homebrew (continue-on-error), so verify the
# asset is present before advertising a release the site can't actually serve.
- name: Verify website release assets
if: ${{ steps.guard.outputs.enabled == 'true' }}
env:
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
VERSION: ${{ needs.release-please.outputs.version }}
TAG: ${{ needs.release-please.outputs.tag_name }}
run: |
set -euo pipefail
gh release view "$TAG" --repo "$GITHUB_REPOSITORY" --json assets \
--jq '.assets[].name' > /tmp/release-assets
for asset in \
"gitlawb-node-${VERSION}-x86_64-pc-windows-msvc.zip" \
"gitlawb-node-${VERSION}-x86_64-pc-windows-msvc.zip.sha256"; do
grep -Fxq "$asset" /tmp/release-assets || {
echo "::error::release asset missing: $asset (Windows build likely failed); not syncing web"
exit 1
}
done
- name: Sync and open PR
if: ${{ steps.guard.outputs.enabled == 'true' }}
env:
GH_TOKEN: ${{ secrets.WEB_SYNC_PAT }}
VERSION: ${{ needs.release-please.outputs.version }}
TAG: ${{ needs.release-please.outputs.tag_name }}
run: |
set -euo pipefail
# Canonical install scripts (source of truth lives in the node repo).
cp node/install.sh web/public/install.sh
cp node/install.ps1 web/public/install.ps1
# Single version source the site can read.
printf '{\n "version": "%s",\n "tag": "%s"\n}\n' "$VERSION" "$TAG" > web/public/version.json
# One-time, idempotent drift fixes: drop orphaned binaries, fix stale link.
rm -rf web/public/bin
if [ -f web/public/skill.md ]; then
sed -i 's#github.com/gitlawb/releases#github.com/Gitlawb/node/releases#g' web/public/skill.md
fi
cd web
# Stage first so brand-new (untracked) files are caught by the no-op check.
git add -A
if git diff --cached --quiet; then
echo "web already up to date."
exit 0
fi
branch="release-sync/${TAG}"
git config user.name "github-actions[bot]"
git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
git checkout -b "$branch"
git commit -m "chore: sync gitlawb ${TAG} (install scripts + version)"
# Credentials are not persisted in .git/config; supply the token only for the push.
git push --force "https://x-access-token:${GH_TOKEN}@github.com/Gitlawb/web.git" "HEAD:$branch"
gh pr create --repo Gitlawb/web --head "$branch" \
--title "Sync gitlawb ${TAG}" \
--body "Automated sync from Gitlawb/node ${TAG}: install.sh, install.ps1, public/version.json, and one-time drift fixes (removed orphaned public/bin, fixed skill.md releases link). Review and merge to deploy via Vercel." \
|| {
# Only swallow the "PR already exists" case; surface auth/API failures
# (a failed gh pr list must not be misread as "PR exists").
pr_count="$(gh pr list --repo Gitlawb/web --head "$branch" --state open --json number --jq 'length')" || exit 1
if [ "$pr_count" != "0" ]; then
echo "PR for $branch already exists; branch was updated."
else
exit 1
fi
}
+8 -7
View File
@@ -23,17 +23,18 @@ jobs:
# Act on PRs only; never on issues.
days-before-issue-stale: -1
days-before-issue-close: -1
# Only PRs the triage step flagged as needing an author response.
only-pr-labels: "needs-issue,needs-description"
# Only PRs the triage step flagged as missing a real description.
# needs-issue is advisory-only (linking an issue is a SHOULD, and many
# legitimate small PRs have none) so it does not drive auto-close.
only-pr-labels: "needs-description"
stale-pr-label: "stale"
exempt-pr-labels: "workflow-change,security,pinned"
remove-stale-when-updated: true
stale-pr-message: >
This PR has been inactive for 14 days and is still missing a linked issue
or a description. It will be closed in 7 days if there's no update. Push a
change or leave a comment to keep it open — no hard feelings, you can reopen
anytime.
This PR has been inactive for 14 days and is still missing a description.
It will be closed in 7 days if there's no update. Push a change or leave a
comment to keep it open. No hard feelings, you can reopen anytime.
close-pr-message: >
Closing due to inactivity. Reopen whenever you're ready to update it — see
Closing due to inactivity. Reopen whenever you're ready to update it. See
CONTRIBUTING.md for what helps us review quickly.
ascending: true
+1 -1
View File
@@ -1,3 +1,3 @@
{
".": "0.3.8"
".": "0.4.0"
}
+52
View File
@@ -0,0 +1,52 @@
# Changelog
## [0.4.0](https://github.com/Gitlawb/node/compare/v0.3.9...v0.4.0) (2026-06-30)
### Features
* agent profiles (display name, bio, avatar, social links) ([#23](https://github.com/Gitlawb/node/issues/23)) ([09a3397](https://github.com/Gitlawb/node/commit/09a339745eca40a2567d911e947e3fc7426fc621))
* **db:** versioned schema migrations with idempotent backfill ([#21](https://github.com/Gitlawb/node/issues/21)) ([927e4d0](https://github.com/Gitlawb/node/commit/927e4d0cfb4ea11dca6930780939afaa067797ba))
* encrypted replication for private subtrees (B1/B2/B3) for [#18](https://github.com/Gitlawb/node/issues/18) ([#36](https://github.com/Gitlawb/node/issues/36)) ([5ff7af8](https://github.com/Gitlawb/node/commit/5ff7af84fa21fab53a12dfa04a0e7fb7e7d672e6))
* **git-remote-gitlawb:** add --version and --help flags ([#30](https://github.com/Gitlawb/node/issues/30)) ([3a401eb](https://github.com/Gitlawb/node/commit/3a401eb0792f9c5f5a10de6c16861655ab3836e0))
* **gitlawb-attest:** External Attestation v1 for ref-update certs ([#20](https://github.com/Gitlawb/node/issues/20)) ([924bccd](https://github.com/Gitlawb/node/commit/924bccd8e53e9be2dc9d6d4e4f1376952d6462bb))
* **node:** blind recipient identities at rest and gate B1 by repo readability ([#40](https://github.com/Gitlawb/node/issues/40)) ([abdc775](https://github.com/Gitlawb/node/commit/abdc7757708d2bbb2bfda99bb65d50756142a42e))
* **node:** enforce per-route authorization across the REST and GraphQL surface ([#87](https://github.com/Gitlawb/node/issues/87)) ([2202b00](https://github.com/Gitlawb/node/commit/2202b0097fab6976ab366a2cdc385f1146a72f86))
* **node:** graceful shutdown + Prometheus metrics endpoint ([#22](https://github.com/Gitlawb/node/issues/22)) ([2ce4da9](https://github.com/Gitlawb/node/commit/2ce4da9cc5a4791e8ae5a1cd90270b67f60d9ec3))
* **node:** iCaptcha proof-of-intelligence gate on create_repo + register ([#108](https://github.com/Gitlawb/node/issues/108)) ([adc20f9](https://github.com/Gitlawb/node/commit/adc20f9effad7b42fab55002875d209c4ed79518))
* **node:** iCaptcha-aware repo propagation gate with quarantine ([#125](https://github.com/Gitlawb/node/issues/125)) ([8b9ceec](https://github.com/Gitlawb/node/commit/8b9ceec25ef338965d1db72d39a7f2adb5300cc9))
* **node:** owner-only push enforcement behind GITLAWB_ENFORCE_OWNER_PUSH ([#31](https://github.com/Gitlawb/node/issues/31)) ([#68](https://github.com/Gitlawb/node/issues/68)) ([0a15e76](https://github.com/Gitlawb/node/commit/0a15e763d2ab46737a3831715facbe51045b33ba))
* **node:** peer partial-mirrors for repos with private subtrees ([#35](https://github.com/Gitlawb/node/issues/35)) ([e365a57](https://github.com/Gitlawb/node/commit/e365a57b51c167abf06e87bfeb0565c80ed1b849))
* **node:** pin the per-push object delta instead of re-enumerating the whole repo ([#90](https://github.com/Gitlawb/node/issues/90)) ([1af4fdf](https://github.com/Gitlawb/node/commit/1af4fdf485c7084cd160551b757b1ad1eed65cc6))
* **node:** replication enforcement (Phase 2) for [#18](https://github.com/Gitlawb/node/issues/18) ([#34](https://github.com/Gitlawb/node/issues/34)) ([8680d0f](https://github.com/Gitlawb/node/commit/8680d0f9d6600bba1a52d15624f8a2802a169511))
* **node:** signature-gated agent self-deregister ([#29](https://github.com/Gitlawb/node/issues/29)) ([#63](https://github.com/Gitlawb/node/issues/63)) ([ff492b4](https://github.com/Gitlawb/node/commit/ff492b452126f5568dac4286a7249f0cadb8b380))
* **node:** subtree content withholding (Phase 3) for [#18](https://github.com/Gitlawb/node/issues/18) ([#28](https://github.com/Gitlawb/node/issues/28)) ([61b3830](https://github.com/Gitlawb/node/commit/61b383019fd895a1a6adfad934ad6c626e0f095e))
* path-scoped repository visibility (Phase 1) for [#18](https://github.com/Gitlawb/node/issues/18) ([#25](https://github.com/Gitlawb/node/issues/25)) ([6abaf1d](https://github.com/Gitlawb/node/commit/6abaf1d7ed8fc55c6547568ae7247131311bde98))
* per-DID rate limiting on creation endpoints (10/hour) ([#13](https://github.com/Gitlawb/node/issues/13)) ([b12c6bc](https://github.com/Gitlawb/node/commit/b12c6bc3283c2647224a62fb520a3cb7acf4a747))
* **sync:** auto-register as replica with origin after successful mirror ([#56](https://github.com/Gitlawb/node/issues/56)) ([c03c9af](https://github.com/Gitlawb/node/commit/c03c9af8fadafe262a2bf3cf25e19edf7160d376))
### Bug Fixes
* **api:** blob endpoint returns 400/404 instead of 500 on bad paths ([#37](https://github.com/Gitlawb/node/issues/37)) ([b61a1bd](https://github.com/Gitlawb/node/commit/b61a1bd46ffe78b39c0967e97b0ad349eba0b046))
* **core:** route seed access through the zeroizing wrapper ([#41](https://github.com/Gitlawb/node/issues/41)) ([#64](https://github.com/Gitlawb/node/issues/64)) ([c9f43b0](https://github.com/Gitlawb/node/commit/c9f43b010576edb3c92a3be3d935cad232250344))
* **core:** zeroize the derived X25519 secret ([#65](https://github.com/Gitlawb/node/issues/65)) ([#91](https://github.com/Gitlawb/node/issues/91)) ([2f6611a](https://github.com/Gitlawb/node/commit/2f6611a30fbb66cef9991cf4c6e507548acc5038))
* **gl:** paginate gl-clone Arweave recovery and make /encrypted-blobs parsing schema-strict ([#49](https://github.com/Gitlawb/node/issues/49)) ([#70](https://github.com/Gitlawb/node/issues/70)) ([2153b0b](https://github.com/Gitlawb/node/commit/2153b0b7a67a48c29a33cd70b80cfbb69760805d))
* **infra:** drop fly idle_timeout 600 -&gt; 120 ([#38](https://github.com/Gitlawb/node/issues/38)) ([a2217bf](https://github.com/Gitlawb/node/commit/a2217bff27ff3d6cae02bd8a12f8a0af7ce2b0a1))
* **node:** anchor the real old_sha and issue a per-ref certificate ([#72](https://github.com/Gitlawb/node/issues/72)) ([6809201](https://github.com/Gitlawb/node/commit/6809201daa223d6f833be6e95876a7b4e1f2b0b5))
* **node:** close under-withholding via full ref scope and full-history classification ([#42](https://github.com/Gitlawb/node/issues/42)) ([#84](https://github.com/Gitlawb/node/issues/84)) ([3e1e904](https://github.com/Gitlawb/node/commit/3e1e9045e4e3aa5ea0aee69767ceb01637920d2a))
* **node:** dedupe mirror and canonical repo rows on list surfaces ([#6](https://github.com/Gitlawb/node/issues/6)) ([#73](https://github.com/Gitlawb/node/issues/73)) ([3e8e333](https://github.com/Gitlawb/node/commit/3e8e333aa03d7a2fe455d5d83f23089d58feb8c9))
* **node:** enforce path-scoped visibility on the REST read API ([#52](https://github.com/Gitlawb/node/issues/52)) ([e37ea7f](https://github.com/Gitlawb/node/commit/e37ea7fec6d5a3171b526c84f884670bbbd258fb))
* **node:** fail closed when a recipient DID can't be resolved ([#47](https://github.com/Gitlawb/node/issues/47)) ([#67](https://github.com/Gitlawb/node/issues/67)) ([abc9ad0](https://github.com/Gitlawb/node/commit/abc9ad03acb48e608234650a25049622673fa53a))
* **node:** gate fork_repo on per-caller path-scoped visibility ([#98](https://github.com/Gitlawb/node/issues/98)) ([#109](https://github.com/Gitlawb/node/issues/109)) ([6ae316c](https://github.com/Gitlawb/node/commit/6ae316cc88521747cbacb2a612b9433897d2e490))
* **node:** gate GET /ipfs/{cid} on per-caller path-scoped visibility ([#110](https://github.com/Gitlawb/node/issues/110)) ([#128](https://github.com/Gitlawb/node/issues/128)) ([174f25a](https://github.com/Gitlawb/node/commit/174f25a206380b26796b8782e1bd860b0a409fc9))
* **node:** gate repo-listing and stats surfaces on visibility ([#97](https://github.com/Gitlawb/node/issues/97), [#99](https://github.com/Gitlawb/node/issues/99), [#101](https://github.com/Gitlawb/node/issues/101), [#104](https://github.com/Gitlawb/node/issues/104)) ([#111](https://github.com/Gitlawb/node/issues/111)) ([828dd27](https://github.com/Gitlawb/node/commit/828dd279a286f58bbb3c73627b2a1e23778b25cf))
* **node:** make Tigris repo hydration resilient to corrupt archives & failed writes ([#54](https://github.com/Gitlawb/node/issues/54)) ([7a99d0f](https://github.com/Gitlawb/node/commit/7a99d0f27cfa869e9f8a803f16cff30191745a51))
* **node:** preserve promisor mirror mode on unknown withheld-paths lookup ([#48](https://github.com/Gitlawb/node/issues/48)) ([#69](https://github.com/Gitlawb/node/issues/69)) ([96fcfdb](https://github.com/Gitlawb/node/commit/96fcfdb1ca6d4cd27226670068702c9f177e283a))
* **node:** reap leaked git child processes ([#53](https://github.com/Gitlawb/node/issues/53)) ([#61](https://github.com/Gitlawb/node/issues/61)) ([803d83e](https://github.com/Gitlawb/node/commit/803d83efdbcfa9d98affac1c5aaf8aacc511ae64))
* **node:** reject malformed path globs with non-trailing or empty-segment wildcards ([#74](https://github.com/Gitlawb/node/issues/74)) ([#75](https://github.com/Gitlawb/node/issues/75)) ([3d880cd](https://github.com/Gitlawb/node/commit/3d880cd56fed10e4c5ae787b52ce411384950f5e))
* **node:** skip withheld-walk when no path-scoped rule can withhold ([#60](https://github.com/Gitlawb/node/issues/60)) ([338ff83](https://github.com/Gitlawb/node/commit/338ff83584f2ab960be7e42ecec64191f5aaeb95))
* **release:** give crates explicit versions for release-please ([#129](https://github.com/Gitlawb/node/issues/129)) ([788a868](https://github.com/Gitlawb/node/commit/788a8686c1deafe03573703d72eb927cde81f54d))
* **release:** use simple release-type + generic Cargo.toml updaters ([#130](https://github.com/Gitlawb/node/issues/130)) ([12bfb1b](https://github.com/Gitlawb/node/commit/12bfb1b3c86b38f0b22d488cd6e85dfd96e0c37f))
* **security:** gate webhook creation through the public-host validator ([#81](https://github.com/Gitlawb/node/issues/81)) ([#92](https://github.com/Gitlawb/node/issues/92)) ([f28fa02](https://github.com/Gitlawb/node/commit/f28fa02236bd65c6a4fb690131ab14640cc53a12))
* **security:** reject non-public peer URLs + prune poisoned peers ([#78](https://github.com/Gitlawb/node/issues/78)) ([a8cc33a](https://github.com/Gitlawb/node/commit/a8cc33a185f2649d3fe100ec271ee5739a55eba7))
+4 -3
View File
@@ -69,9 +69,10 @@ expected to clear a basic quality bar:
- **A real description.** Say what changes and why. "Update code" is not a description.
A triage bot labels PRs that are missing these and leaves a short note. Nothing is closed
automatically while you're engaging. A flagged PR that goes 14 days with no linked issue or
description gets a stale warning, and is closed 7 days later if still untouched. Closed PRs can
be reopened at any time once updated.
automatically while you're engaging. A flagged PR still missing a description after 14 days of
inactivity gets a stale warning, and is closed 7 days later if still untouched. A missing issue
link alone never triggers closure; that label stays advisory. Closed PRs can be reopened at any
time once updated.
## Development environment
Generated
+22 -19
View File
@@ -329,7 +329,7 @@ dependencies = [
"keccak-asm",
"paste",
"proptest",
"rand 0.9.2",
"rand 0.9.3",
"rapidhash",
"ruint",
"rustc-hash",
@@ -2107,7 +2107,7 @@ version = "3.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "faf9468729b8cbcea668e36183cb69d317348c2e08e994829fb56ebfdfbaac34"
dependencies = [
"windows-sys 0.61.2",
"windows-sys 0.48.0",
]
[[package]]
@@ -2262,7 +2262,7 @@ checksum = "6ddc2d09feefeee8bd78101665bd8645637828fa9317f9f292496dbbd8c65ff3"
dependencies = [
"crc",
"digest 0.10.7",
"rand 0.9.2",
"rand 0.9.3",
"regex",
"rustversion",
]
@@ -2928,7 +2928,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb"
dependencies = [
"libc",
"windows-sys 0.61.2",
"windows-sys 0.52.0",
]
[[package]]
@@ -3365,6 +3365,7 @@ dependencies = [
"aws-config",
"aws-sdk-s3",
"axum",
"base64",
"bytes",
"chrono",
"cid",
@@ -3376,6 +3377,7 @@ dependencies = [
"hex",
"hmac",
"http-body-util",
"libc",
"libp2p-core",
"libp2p-dns",
"libp2p-gossipsub",
@@ -3402,6 +3404,7 @@ dependencies = [
"tower-http",
"tracing",
"tracing-subscriber",
"unicode-normalization",
"uuid",
"zstd",
]
@@ -3598,7 +3601,7 @@ dependencies = [
"idna",
"ipnet",
"once_cell",
"rand 0.9.2",
"rand 0.9.3",
"ring",
"thiserror 2.0.18",
"tinyvec",
@@ -3620,7 +3623,7 @@ dependencies = [
"moka",
"once_cell",
"parking_lot",
"rand 0.9.2",
"rand 0.9.3",
"resolv-conf",
"smallvec",
"thiserror 2.0.18",
@@ -4535,7 +4538,7 @@ dependencies = [
"hyper-util",
"log",
"pin-project-lite",
"rand 0.9.2",
"rand 0.9.3",
"regex",
"serde_json",
"serde_urlencoded",
@@ -5282,7 +5285,7 @@ dependencies = [
"bit-vec",
"bitflags 2.11.0",
"num-traits",
"rand 0.9.2",
"rand 0.9.3",
"rand_chacha 0.9.0",
"rand_xorshift",
"regex-syntax",
@@ -5342,14 +5345,14 @@ dependencies = [
[[package]]
name = "quinn-proto"
version = "0.11.14"
version = "0.11.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "434b42fec591c96ef50e21e886936e66d3cc3f737104fdb9b737c40ffb94c098"
checksum = "4fcb935c5bec503c2f0e306bdd3e58bb9029dcb14fa8d9ac76e3a5256ac0763e"
dependencies = [
"bytes",
"getrandom 0.3.4",
"lru-slab",
"rand 0.9.2",
"rand 0.9.3",
"ring",
"rustc-hash",
"rustls",
@@ -5416,9 +5419,9 @@ dependencies = [
[[package]]
name = "rand"
version = "0.9.2"
version = "0.9.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6db2770f06117d490610c7488547d543617b21bfa07796d7a12f6f1bd53850d1"
checksum = "7ec095654a25171c2124e9e3393a930bddbffdc939556c914957a4c3e0a87166"
dependencies = [
"rand_chacha 0.9.0",
"rand_core 0.9.5",
@@ -5738,7 +5741,7 @@ dependencies = [
"primitive-types",
"proptest",
"rand 0.8.6",
"rand 0.9.2",
"rand 0.9.3",
"rlp",
"ruint-macro",
"serde_core",
@@ -5801,7 +5804,7 @@ dependencies = [
"errno",
"libc",
"linux-raw-sys",
"windows-sys 0.61.2",
"windows-sys 0.52.0",
]
[[package]]
@@ -6676,9 +6679,9 @@ checksum = "55937e1799185b12863d447f42597ed69d9928686b8d88a1df17376a097d8369"
[[package]]
name = "tar"
version = "0.4.45"
version = "0.4.46"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "22692a6476a21fa75fdfc11d452fda482af402c008cdbaf3476414e122040973"
checksum = "3f6221d9a6003c78398e3b239969f352578258df48c8eb051caadae0015bc840"
dependencies = [
"filetime",
"libc",
@@ -6695,7 +6698,7 @@ dependencies = [
"getrandom 0.4.2",
"once_cell",
"rustix",
"windows-sys 0.61.2",
"windows-sys 0.52.0",
]
[[package]]
@@ -7058,7 +7061,7 @@ dependencies = [
"http 1.4.0",
"httparse",
"log",
"rand 0.9.2",
"rand 0.9.3",
"sha1",
"thiserror 2.0.18",
"utf-8",
+11 -1
View File
@@ -9,7 +9,13 @@ members = [
]
[workspace.package]
version = "0.3.9"
# Per-crate versions live in each crate's [package] so release-please can manage
# them (it cannot rewrite an inherited `version.workspace = true`). They are kept
# in lockstep by release-please's generic updater: each crate's `version` line
# carries an `# x-release-please-version` annotation and is listed under
# `extra-files` in release-please-config.json. The release type is `simple`
# (not `rust`) because the rust strategy unconditionally tries to rewrite this
# virtual workspace root as a package manifest and errors out.
edition = "2021"
rust-version = "1.91"
license = "MIT OR Apache-2.0"
@@ -48,3 +54,7 @@ uuid = { version = "1", features = ["v4"] }
reqwest = { version = "0.12", features = ["blocking", "json", "multipart", "rustls-tls"], default-features = false }
# HMAC
hmac = "0.12"
[profile.release]
lto = "thin"
strip = "symbols"
+9 -1
View File
@@ -44,6 +44,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
git \
ca-certificates \
curl \
tini \
&& rm -rf /var/lib/apt/lists/*
# Non-root user for runtime
@@ -72,4 +73,11 @@ VOLUME ["/data"]
HEALTHCHECK --interval=30s --timeout=5s --start-period=15s --retries=3 \
CMD curl -fsS http://127.0.0.1:${GITLAWB_PORT}/health || exit 1
ENTRYPOINT ["gitlawb-node"]
# Run under tini so the node is never PID 1. As PID 1 a process must reap
# reparented orphans itself; the node does not, so git's reparented grandchildren
# (e.g. pack-objects orphaned when a served upload-pack dies on a client
# disconnect) would accumulate as zombies until fork() fails with EAGAIN (#53).
# tini reaps them and forwards SIGTERM/SIGINT to the node, which runs its own
# graceful shutdown. (No `-g`: the node manages its own children, so group-wide
# signalling would only disturb in-flight git helpers during shutdown.)
ENTRYPOINT ["tini", "--", "gitlawb-node"]
+10 -2
View File
@@ -119,10 +119,18 @@ docker compose down
## Install the CLI
Use the installer once releases are available for your platform:
```bash
# npm (macOS / Linux)
npm install -g @gitlawb/gl
# Homebrew (macOS / Linux)
brew install gitlawb/tap/gl
# curl (macOS / Linux)
curl -fsSL https://gitlawb.com/install.sh | sh
# PowerShell (Windows)
irm https://gitlawb.com/install.ps1 | iex
```
Or build from source:
+1 -1
View File
@@ -1,7 +1,7 @@
[package]
name = "git-remote-gitlawb"
description = "Git remote helper — enables 'git clone gitlawb://did:gitlawb:...'"
version.workspace = true
version = "0.4.0" # x-release-please-version
edition.workspace = true
rust-version.workspace = true
license.workspace = true
+7 -3
View File
@@ -272,8 +272,7 @@ fn handle_connect(
let mut req = client
.post(&post_url)
.header("Content-Type", format!("application/x-{}-request", service))
.header("User-Agent", "git/2.0 git-remote-gitlawb/0.1.0")
.body(request_body.clone());
.header("User-Agent", "git/2.0 git-remote-gitlawb/0.1.0");
// Add RFC 9421 HTTP Signature auth on push operations
if service == "git-receive-pack" {
@@ -291,7 +290,12 @@ fn handle_connect(
}
}
let pack_resp = req.send().with_context(|| format!("POST {post_url}"))?;
// Attach the body after signing so the pack bytes are moved, not cloned —
// packs can be large and the clone doubled peak memory on push.
let pack_resp = req
.body(request_body)
.send()
.with_context(|| format!("POST {post_url}"))?;
if !pack_resp.status().is_success() {
bail!("POST /{} returned {}", service, pack_resp.status());
+1 -1
View File
@@ -1,7 +1,7 @@
[package]
name = "gitlawb-attest"
description = "External Attestation v1: pluggable provenance attachments for gitlawb ref-update certs"
version.workspace = true
version = "0.4.0" # x-release-please-version
edition.workspace = true
rust-version.workspace = true
license.workspace = true
+1 -1
View File
@@ -127,7 +127,7 @@ impl Attestation {
self.type_
)));
}
Ok(serde_json::from_value(self.payload.clone())?)
Ok(P::deserialize(&self.payload)?)
}
}
+1 -1
View File
@@ -1,7 +1,7 @@
[package]
name = "gitlawb-core"
description = "Core cryptographic primitives for the gitlawb network: DIDs, CIDs, UCAN, HTTP Signatures, ref-update certificates"
version.workspace = true
version = "0.4.0" # x-release-please-version
edition.workspace = true
rust-version.workspace = true
license.workspace = true
+11 -6
View File
@@ -6,6 +6,7 @@
use crate::identity::Keypair;
use anyhow::{Context, Result};
use ed25519_dalek::VerifyingKey;
use zeroize::Zeroizing;
/// X25519 public key (Montgomery u) for an Ed25519 verifying key.
fn x25519_public(vk: &VerifyingKey) -> Result<[u8; 32]> {
@@ -18,14 +19,18 @@ fn x25519_public(vk: &VerifyingKey) -> Result<[u8; 32]> {
}
/// X25519 secret scalar for an Ed25519 seed (SHA-512 of seed, lower 32, clamped).
fn x25519_secret_from_seed(seed: &[u8; 32]) -> [u8; 32] {
/// Returns the scalar wrapped in `Zeroizing`, and scrubs the intermediate
/// SHA-512 digest, so no copy of this secret material lingers in freed memory.
fn x25519_secret_from_seed(seed: &[u8; 32]) -> Zeroizing<[u8; 32]> {
use sha2::{Digest, Sha512};
let h = Sha512::digest(seed);
let mut s = [0u8; 32];
use zeroize::Zeroize;
let mut h = Sha512::digest(seed);
let mut s = Zeroizing::new([0u8; 32]);
s.copy_from_slice(&h[..32]);
s[0] &= 248;
s[31] &= 127;
s[31] |= 64;
h.as_mut_slice().zeroize();
s
}
@@ -123,7 +128,7 @@ pub fn open_blob(envelope: &[u8], keypair: &Keypair) -> Result<Vec<u8>> {
.context("decode header")?;
let body = &envelope[p + hlen..];
let my_x = XSecret::from(x25519_secret_from_seed(&keypair.seed_bytes()));
let my_x = XSecret::from(*x25519_secret_from_seed(&keypair.to_seed()));
// Identities are blinded: no entry says which recipient it belongs to, so
// try each one. The ChaChaBox AEAD tag authenticates, so exactly the
@@ -185,10 +190,10 @@ mod tests {
// The X25519 public derived from the Ed25519 public must equal the
// X25519 public of the X25519 secret derived from the same seed.
let kp = Keypair::generate();
let seed = kp.seed_bytes();
let seed = kp.to_seed();
let xpub_from_public = x25519_public(&kp.verifying_key()).unwrap();
let xsec = x25519_secret_from_seed(&seed);
let xpub_from_secret = crypto_box::SecretKey::from(xsec).public_key().to_bytes();
let xpub_from_secret = crypto_box::SecretKey::from(*xsec).public_key().to_bytes();
assert_eq!(xpub_from_public, xpub_from_secret);
}
+1 -1
View File
@@ -129,7 +129,7 @@ impl HttpSignature {
pub fn missing_components(&self) -> Vec<&str> {
COVERED_COMPONENTS
.iter()
.filter(|c| !self.components.contains(&c.to_string()))
.filter(|c| !self.components.iter().any(|s| s.as_str() == **c))
.copied()
.collect()
}
+17 -7
View File
@@ -52,13 +52,9 @@ impl Keypair {
URL_SAFE_NO_PAD.encode(sig.to_bytes())
}
/// The raw 32-byte Ed25519 seed. Used to derive the X25519 secret for
/// envelope decryption (see `crate::encrypt`).
pub fn seed_bytes(&self) -> [u8; 32] {
self.signing_key.to_bytes()
}
/// Export the signing key as raw 32-byte seed (wrapped in Zeroizing).
/// The raw 32-byte Ed25519 seed, wrapped in `Zeroizing` so the copy is
/// scrubbed on drop. This is the only seed accessor; it is used to derive
/// the X25519 secret for envelope decryption (see `crate::encrypt`).
pub fn to_seed(&self) -> Zeroizing<[u8; 32]> {
Zeroizing::new(self.signing_key.to_bytes())
}
@@ -172,6 +168,20 @@ mod tests {
assert_eq!(kp.verifying_key(), kp2.verifying_key());
}
// `to_seed()` is the only seed accessor on `Keypair` (the raw, unzeroized
// `seed_bytes()` was removed in #41). Pin its return type so a bare-array
// accessor can't slip back in, and confirm it is deterministic. The
// from_seed -> verifying_key round-trip is already covered by
// `from_seed_round_trip`.
#[test]
fn to_seed_returns_zeroizing_and_is_deterministic() {
let kp = Keypair::generate();
// The type annotation is the regression guard: the seed must be handed
// out wrapped in `Zeroizing`, never as a bare `[u8; 32]`.
let seed: Zeroizing<[u8; 32]> = kp.to_seed();
assert_eq!(*kp.to_seed(), *seed); // stable across calls
}
#[test]
fn sign_b64_decodes_to_valid_signature() {
use base64::{engine::general_purpose::URL_SAFE_NO_PAD, Engine};
+7 -2
View File
@@ -1,7 +1,7 @@
[package]
name = "gitlawb-node"
description = "The gitlawb node daemon — git hosting over HTTP with DID auth"
version.workspace = true
version = "0.4.0" # x-release-please-version
edition.workspace = true
rust-version.workspace = true
license.workspace = true
@@ -13,6 +13,7 @@ path = "src/main.rs"
[dependencies]
gitlawb-core = { path = "../gitlawb-core" }
ed25519-dalek = { workspace = true }
base64 = { workspace = true }
tokio = { workspace = true }
serde = { workspace = true }
serde_json = { workspace = true }
@@ -28,9 +29,10 @@ async-graphql-axum = "7"
tokio-stream = { version = "0.1", features = ["sync"] }
tower = "0.5"
tower-http = { version = "0.6", features = ["cors", "trace", "request-id", "limit"] }
sqlx = { version = "0.8", features = ["postgres", "runtime-tokio-rustls", "chrono", "uuid"] }
sqlx = { version = "0.8", features = ["postgres", "runtime-tokio-rustls", "chrono", "uuid", "macros", "migrate"] }
clap = { version = "4", features = ["derive", "env"] }
bytes = "1"
libc = "0.2"
cid = { workspace = true }
hex = { workspace = true }
sha2 = { workspace = true }
@@ -38,6 +40,9 @@ hmac = { workspace = true }
http-body-util = "0.1"
tokio-util = { version = "0.7", features = ["io"] }
dirs-next = "2"
# NFC-normalize visibility globs/paths at the matcher seam so a deny rule
# withholds a canonically-equivalent path regardless of NFC/NFD form (#101).
unicode-normalization = "0.1"
reqwest = { workspace = true }
libp2p-core = "0.43"
libp2p-gossipsub = "0.49.4"
+103 -2
View File
@@ -2,9 +2,10 @@
use axum::extract::{Path, Query, State};
use axum::http::StatusCode;
use axum::Json;
use axum::{Extension, Json};
use serde::{Deserialize, Serialize};
use crate::auth::AuthenticatedDid;
use crate::error::{AppError, Result};
use crate::state::AppState;
@@ -20,6 +21,14 @@ fn agent_key_segment(did: &str) -> &str {
did.split(':').next_back().unwrap_or(did)
}
/// Whether `caller` (the verified `AuthenticatedDid`) is the same identity as
/// `target` (the DID being acted on), tolerant of full `did:key:...` vs short
/// key-only forms on either side. Used to gate self-deregistration so a DID
/// can only retire itself.
fn caller_matches_did(caller: &str, target: &str) -> bool {
agent_key_segment(caller) == agent_key_segment(target)
}
async fn resolve_agent_did(state: &AppState, did: &str) -> Result<String> {
let normalized_did = normalize_agent_did(did);
if state.db.get_agent(&normalized_did).await?.is_some() {
@@ -71,6 +80,8 @@ pub struct AgentResponse {
pub capabilities: Vec<String>,
pub registered_at: String,
pub last_seen: Option<String>,
/// Lifecycle status: `active` or `revoked`.
pub status: String,
}
/// GET /api/v1/agents
@@ -87,6 +98,7 @@ pub async fn list_agents(
capabilities: a.capabilities,
registered_at: a.registered_at,
last_seen: a.last_seen,
status: a.status,
})
.collect();
Ok(Json(serde_json::json!({ "agents": list })))
@@ -111,6 +123,7 @@ pub async fn show_agent(
capabilities: agent.capabilities,
registered_at: agent.registered_at,
last_seen: agent.last_seen,
status: agent.status,
}),
))
}
@@ -133,9 +146,97 @@ pub async fn get_trust(
}))
}
/// DELETE /api/v1/agents/{did}
///
/// Self-deregistration: the holder of a DID's key marks their own agent
/// `revoked`, removing it from discovery (issue #29). Authenticated by the
/// rfc9421 signature middleware; a caller may only retire its own DID.
pub async fn deregister_agent(
State(state): State<AppState>,
Extension(auth): Extension<AuthenticatedDid>,
Path(did): Path<String>,
) -> Result<StatusCode> {
// Authorize on the verified identity, and act on it too. The path DID is
// only an intent check (it must name the caller); the row we revoke is the
// authenticated DID itself, never a value derived from the untrusted path.
// Revoking `resolve_agent_did(did)` instead would let the fuzzy prefix
// resolver act on a different identity than the one just authorized.
if !caller_matches_did(&auth.0, &did) {
return Err(AppError::BadRequest(
"an agent can only deregister itself".into(),
));
}
if state.db.revoke_agent(&auth.0).await? {
Ok(StatusCode::NO_CONTENT)
} else {
Err(AppError::NotFound(format!("agent {} not found", auth.0)))
}
}
#[cfg(test)]
mod tests {
use super::{agent_key_segment, normalize_agent_did};
use super::{agent_key_segment, caller_matches_did, normalize_agent_did, AgentResponse};
#[test]
fn caller_matches_own_did_full_form() {
assert!(caller_matches_did(
"did:key:z6MkExample",
"did:key:z6MkExample"
));
}
#[test]
fn caller_matches_own_did_short_form() {
// Authenticated full DID vs short path form, and vice versa.
assert!(caller_matches_did("did:key:z6MkExample", "z6MkExample"));
assert!(caller_matches_did("z6MkExample", "did:key:z6MkExample"));
}
#[test]
fn caller_cannot_revoke_a_different_did() {
// The core authorization property for issue #29.
assert!(!caller_matches_did(
"did:key:z6MkAttacker",
"did:key:z6MkVictim"
));
assert!(!caller_matches_did("did:key:z6MkAttacker", "z6MkVictim"));
}
#[test]
fn agent_response_surfaces_status() {
// A revoked DID's response must carry its status so callers can see it
// is retired (issue #29).
let resp = AgentResponse {
did: "did:key:orphan".to_string(),
trust_score: 0.1,
capabilities: vec!["reputation:score".to_string()],
registered_at: "2026-06-19T00:00:00Z".to_string(),
last_seen: None,
status: "revoked".to_string(),
};
let json = serde_json::to_value(&resp).unwrap();
assert_eq!(json["status"], "revoked");
assert!(json.get("replaced_by").is_none());
}
#[test]
fn agent_response_active_status() {
let resp = AgentResponse {
did: "did:key:active".to_string(),
trust_score: 0.5,
capabilities: vec![],
registered_at: "2026-06-19T00:00:00Z".to_string(),
last_seen: None,
status: "active".to_string(),
};
let json = serde_json::to_value(&resp).unwrap();
assert_eq!(json["status"], "active");
}
#[test]
fn normalize_agent_did_preserves_full_did() {
+26 -12
View File
@@ -79,12 +79,9 @@ pub async fn create_bounty(
return Err(AppError::BadRequest("amount must be positive".into()));
}
// Verify repo exists
let _ = state
.db
.get_repo(&owner, &repo)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{repo}")))?;
// Read-gate: anyone who can read the repo may fund a bounty on it (this closes
// the private-repo bounty leak); ownership is not required.
crate::api::authorize_repo_read(&state, &owner, &repo, Some(auth.0.as_str()), "/").await?;
let now = Utc::now().to_rfc3339();
let bounty = BountyRecord {
@@ -213,8 +210,12 @@ pub async fn submit_bounty(
bounty.status
)));
}
if bounty.claimant_did.as_deref() != Some(&auth.0) {
return Err(AppError::BadRequest("only the claimant can submit".into()));
let is_claimant = bounty
.claimant_did
.as_deref()
.is_some_and(|c| crate::api::did_matches(&auth.0, c));
if !is_claimant {
return Err(AppError::Forbidden("only the claimant can submit".into()));
}
let now = Utc::now().to_rfc3339();
@@ -249,8 +250,8 @@ pub async fn approve_bounty(
bounty.status
)));
}
if bounty.creator_did != auth.0 {
return Err(AppError::BadRequest(
if !crate::api::did_matches(&auth.0, &bounty.creator_did) {
return Err(AppError::Forbidden(
"only the bounty creator can approve".into(),
));
}
@@ -296,8 +297,8 @@ pub async fn cancel_bounty(
bounty.status
)));
}
if bounty.creator_did != auth.0 {
return Err(AppError::BadRequest(
if !crate::api::did_matches(&auth.0, &bounty.creator_did) {
return Err(AppError::Forbidden(
"only the bounty creator can cancel".into(),
));
}
@@ -332,6 +333,19 @@ pub async fn dispute_bounty(
)));
}
// Only the bounty creator or the current claimant may dispute (N19 — this
// endpoint had no identity check at all).
let is_creator = crate::api::did_matches(&auth.0, &bounty.creator_did);
let is_claimant = bounty
.claimant_did
.as_deref()
.is_some_and(|c| crate::api::did_matches(&auth.0, c));
if !is_creator && !is_claimant {
return Err(AppError::Forbidden(
"only the bounty creator or claimant can dispute this bounty".into(),
));
}
// Check if deadline exceeded
if let Some(ref claimed_at) = bounty.claimed_at {
if let Ok(claimed) = chrono::DateTime::parse_from_rfc3339(claimed_at) {
+86 -3
View File
@@ -9,28 +9,47 @@
//! `git cat-file <type> <sha256>` (i.e. without the git framing header).
//! This is consistent with how `gitlawb_core::cid::Cid::from_git_object_bytes`
//! computes CIDs when objects are pushed.
//!
//! Serving is access-controlled: an object is returned only from a repo row the
//! requesting caller is permitted to read (per-caller path-scoped visibility,
//! see `get_by_cid`).
use axum::{
extract::{Path, State},
http::{HeaderMap, HeaderName, HeaderValue, StatusCode},
response::{IntoResponse, Response},
Json,
Extension, Json,
};
use cid::CidGeneric;
use std::collections::{HashMap, HashSet};
use std::str::FromStr;
use crate::auth::AuthenticatedDid;
use crate::error::{AppError, Result};
use crate::git::store;
use crate::git::visibility_pack::{has_path_scoped_rule, withheld_blob_oids};
use crate::state::AppState;
use crate::visibility::{visibility_check, Decision};
/// GET /ipfs/{cid}
///
/// Search all repos on the node for a git object whose SHA-256 hash matches
/// the given CIDv1. Returns the raw object content bytes with appropriate
/// headers if found, or 404 if not found.
/// the given CIDv1, returning its raw content if the caller may read it.
///
/// Visibility (#110): the object is served only from a repo row the caller
/// passes. For each iterated row we gate against that row's OWN rules
/// (`visibility_check` at `"/"`), never re-resolving via `authorize_repo_read`
/// — `get_repo`'s fuzzy match could otherwise authorize a different physical
/// row than the one read (KTD2a). When the row carries path-scoped rules, a
/// blob withheld from the caller (`withheld_blob_oids`) is skipped. Denial and
/// genuine not-found both fall through to an opaque 404.
///
/// Scope: this closes the direct unauthenticated scan. A stale-public mirror
/// row still serves withheld content (tracked separately, #124).
pub async fn get_by_cid(
Path(cid_str): Path<String>,
State(state): State<AppState>,
auth: Option<Extension<AuthenticatedDid>>,
) -> Result<Response> {
// 1. Decode the CID and extract the SHA-256 digest
let cid = CidGeneric::<64>::from_str(&cid_str)
@@ -46,6 +65,8 @@ pub async fn get_by_cid(
}
let sha256_hex = hex::encode(mh.digest());
let caller = auth.as_ref().map(|e| e.0 .0.as_str());
let caller_owned = caller.map(|c| c.to_string());
// 2. Search all repos for an object with this SHA-256
let repos = state
@@ -54,12 +75,74 @@ pub async fn get_by_cid(
.await
.map_err(AppError::Internal)?;
// Fetch every repo's visibility rules in one query rather than one per row
// (the gate runs each row against its OWN rules — KTD2a). A row absent from
// the map has no rules.
let repo_ids: Vec<String> = repos.iter().map(|r| r.id.clone()).collect();
let rules_by_repo = state
.db
.list_visibility_rules_for_repos(&repo_ids)
.await
.map_err(AppError::Internal)?;
// Request-scoped memo of the per-repo withheld set (KTD1). The caller is
// constant for one request, so `repo.id` alone is a safe, sufficient key —
// never a coarse caller "class", which `visibility_check`'s exact full-DID
// reader match would make unsafe.
let mut withheld_memo: HashMap<String, HashSet<String>> = HashMap::new();
for repo in &repos {
// Repo-level read gate against THIS row's own rules (KTD2a).
let rules: &[crate::db::VisibilityRule] = rules_by_repo
.get(&repo.id)
.map(Vec::as_slice)
.unwrap_or(&[]);
if visibility_check(rules, repo.is_public, &repo.owner_did, caller, "/") == Decision::Deny {
continue;
}
let repo_path = match state.repo_store.acquire(&repo.owner_did, &repo.name).await {
Ok(p) => p,
Err(_) => continue,
};
// Per-blob withholding only applies when a path-scoped rule exists (KTD4).
if has_path_scoped_rule(rules) {
if !withheld_memo.contains_key(&repo.id) {
let rp = repo_path.clone();
let r = rules.to_vec();
let is_public = repo.is_public;
let owner = repo.owner_did.clone();
let caller_for_walk = caller_owned.clone();
// Full-history walk shells out to git — keep it off the async runtime.
let walk = tokio::task::spawn_blocking(move || {
withheld_blob_oids(&rp, &r, is_public, &owner, caller_for_walk.as_deref())
})
.await;
// Fail closed on EITHER a task panic (JoinError) or a walk error:
// we cannot prove the caller may read here, so skip this repo and
// let a public copy (if any) serve. Never serve on an unproven gate.
let set = match walk {
Ok(Ok(set)) => set,
Ok(Err(e)) => {
tracing::warn!(repo = %repo.name, err = %e, "withheld walk failed; skipping repo");
continue;
}
Err(e) => {
tracing::warn!(repo = %repo.name, err = %e, "withheld walk task panicked; skipping repo");
continue;
}
};
withheld_memo.insert(repo.id.clone(), set);
}
if withheld_memo
.get(&repo.id)
.is_some_and(|set| set.contains(&sha256_hex))
{
continue;
}
}
match store::read_object(&repo_path, &sha256_hex) {
Ok(Some((_obj_type, content))) => {
// 3. Return the content with IPFS-style headers
+36 -11
View File
@@ -39,11 +39,11 @@ pub async fn create_issue(
Path((owner, repo)): Path<(String, String)>,
Json(req): Json<CreateIssueRequest>,
) -> Result<(StatusCode, Json<IssueRecord>)> {
let record = state
.db
.get_repo(&owner, &repo)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{repo}")))?;
// Authorize the caller as a reader before accepting an issue: a non-reader
// must not be able to file an issue against a private repo they cannot read.
// Mirrors create_issue_comment / create_review / create_bounty.
let (record, _rules) =
crate::api::authorize_repo_read(&state, &owner, &repo, Some(auth.0.as_str()), "/").await?;
let issue_id = Uuid::new_v4().to_string();
let now = Utc::now().to_rfc3339();
@@ -161,11 +161,9 @@ pub async fn create_issue_comment(
));
}
let record = state
.db
.get_repo(&owner, &repo)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{repo}")))?;
// Read-gate: a commenter must be able to read the repo, but need not own it.
let (record, _rules) =
crate::api::authorize_repo_read(&state, &owner, &repo, Some(auth.0.as_str()), "/").await?;
let disk_path = state
.repo_store
@@ -207,7 +205,7 @@ pub async fn list_issue_comments(
/// POST /api/v1/repos/{owner}/{repo}/issues/{id}/close
pub async fn close_issue(
State(state): State<AppState>,
Extension(_auth): Extension<AuthenticatedDid>,
Extension(auth): Extension<AuthenticatedDid>,
Path((owner, repo, issue_id)): Path<(String, String, String)>,
) -> Result<Json<serde_json::Value>> {
let record = state
@@ -223,6 +221,33 @@ pub async fn close_issue(
.map_err(|e| AppError::Git(e.to_string()))?;
let disk_path = guard.path().to_path_buf();
// Owner OR issue author may close. The author lives in the issue's git-JSON
// blob (not a DB column); a None author (legacy issues) falls back to
// owner-only. Read it under the write guard, before mutating.
let author_did: Option<String> = match git_issues::get_issue(&disk_path, &issue_id) {
Ok(Some(raw)) => serde_json::from_str::<IssueRecord>(&raw)
.ok()
.and_then(|i| i.author),
Ok(None) => {
guard.release(false).await;
return Err(AppError::NotFound(format!("issue {issue_id} not found")));
}
Err(e) => {
guard.release(false).await;
return Err(AppError::Git(e.to_string()));
}
};
let is_owner = crate::api::require_repo_owner(&record, &auth.0).is_ok();
let is_author = author_did
.as_deref()
.is_some_and(|a| crate::api::did_matches(&auth.0, a));
if !is_owner && !is_author {
guard.release(false).await;
return Err(AppError::Forbidden(
"only the repo owner or the issue author can close this issue".into(),
));
}
let close_result = git_issues::close_issue(&disk_path, &issue_id);
// Always release the advisory lock — even on error; upload to Tigris only on success.
+4 -2
View File
@@ -17,7 +17,7 @@ pub struct LabelRequest {
/// POST /api/v1/repos/:owner/:repo/labels
pub async fn add_label(
State(state): State<AppState>,
Extension(_auth): Extension<AuthenticatedDid>,
Extension(auth): Extension<AuthenticatedDid>,
Path((owner, name)): Path<(String, String)>,
Json(req): Json<LabelRequest>,
) -> Result<(StatusCode, Json<serde_json::Value>)> {
@@ -39,6 +39,7 @@ pub async fn add_label(
.get_repo(&owner, &name)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
crate::api::require_repo_owner(&record, &auth.0)?;
let added = state.db.add_label(&record.id, &label).await?;
let status = if added {
@@ -55,7 +56,7 @@ pub async fn add_label(
/// DELETE /api/v1/repos/:owner/:repo/labels/:label
pub async fn remove_label(
State(state): State<AppState>,
Extension(_auth): Extension<AuthenticatedDid>,
Extension(auth): Extension<AuthenticatedDid>,
Path((owner, name, label)): Path<(String, String, String)>,
) -> Result<Json<serde_json::Value>> {
let record = state
@@ -63,6 +64,7 @@ pub async fn remove_label(
.get_repo(&owner, &name)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
crate::api::require_repo_owner(&record, &auth.0)?;
state.db.remove_label(&record.id, &label).await?;
Ok(Json(serde_json::json!({ "label": label, "removed": true })))
+515
View File
@@ -47,6 +47,13 @@ pub(crate) async fn authorize_repo_read(
.get_repo(owner, name)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
// A quarantined mirror (admitted by the iCaptcha propagation gate but not
// validated) is hidden from every reader — serve/clone and fork alike — as if
// it did not exist, until an operator releases it. Checked before the
// visibility gate so its existence is never disclosed.
if state.db.is_repo_quarantined(&record.id).await? {
return Err(AppError::RepoNotFound(format!("{owner}/{name}")));
}
let rules = state.db.list_visibility_rules(&record.id).await?;
if visibility_check(&rules, record.is_public, &record.owner_did, caller, path) == Decision::Deny
{
@@ -54,3 +61,511 @@ pub(crate) async fn authorize_repo_read(
}
Ok((record, rules))
}
/// Match a presented DID against a stored DID that may be the full `did:key:<id>`
/// form or the bare `<id>` short form (mirror rows store the bare key). Collapse
/// representation only within `did:key`; never let a bare id match across methods —
/// `did:web` / `did:gitlawb` share the base58 space with `did:key`, so a
/// trailing-segment compare would treat `did:key:X` and `did:gitlawb:X` as equal.
pub(crate) fn did_matches(a: &str, b: &str) -> bool {
if a == b {
return true;
}
fn key_id(d: &str) -> &str {
d.strip_prefix("did:key:").unwrap_or(d)
}
let (ka, kb) = (key_id(a), key_id(b));
// After stripping `did:key:`, a value still containing ':' is a non-key full
// DID — do not let it match a bare `did:key` id.
!ka.contains(':') && !kb.contains(':') && ka == kb
}
/// 403 unless `caller` is the repo owner. Uses [`did_matches`] so the owner check
/// and the author check (close policy) share one normalization.
pub(crate) fn require_repo_owner(record: &RepoRecord, caller: &str) -> Result<()> {
if did_matches(caller, &record.owner_did) {
Ok(())
} else {
Err(AppError::Forbidden(
"only the repo owner can perform this action".into(),
))
}
}
#[cfg(test)]
mod did_tests {
use super::did_matches;
#[test]
fn full_matches_bare_same_key() {
assert!(did_matches("did:key:zABC", "zABC"));
assert!(did_matches("zABC", "did:key:zABC"));
}
#[test]
fn rejects_cross_method_collision() {
assert!(!did_matches("did:key:zABC", "did:gitlawb:zABC"));
assert!(!did_matches("did:key:zABC", "did:web:zABC"));
}
#[test]
fn exact_match_and_distinct_keys() {
assert!(did_matches("did:key:zABC", "did:key:zABC"));
assert!(!did_matches("did:key:zABC", "did:key:zXYZ"));
assert!(!did_matches("zABC", "zXYZ"));
}
}
/// Drift guard (plan 002 §Gate-type table, Step 5). Every in-scope mutation
/// handler must contain its expected gate marker in its own body; removing a
/// gate fails this test. Source-level (no DB), so it runs everywhere. When a new
/// route is added to an in-scope group, add its row here with a deliberate gate
/// type — that forced decision is the point.
///
/// Markers are gate-SHAPED — a call (`require_repo_owner(`, `did_matches(`) or a
/// binding/comparison expression (`caller != &record.owner_did`,
/// `let owner_did = auth.0`) — never a bare identifier that could also appear in
/// a log line. Full-line comments are stripped before matching, so a marker that
/// survives only as a comment above a deleted gate does NOT satisfy a row.
#[cfg(test)]
mod authz_guard {
/// The body of `func` with full-line comments removed. Bounds the slice at the
/// next top-level fn item so a marker in a later handler can't leak in,
/// tolerating `pub async`, `pub(crate) async`, `async`, `pub`, and bare `fn`
/// declarations (the old single-`pub async fn` delimiter over-ran on any other
/// form).
fn fn_body(src: &str, func: &str) -> String {
let needle = format!("fn {func}(");
let start = src
.find(&needle)
.unwrap_or_else(|| panic!("handler `{func}` not found (renamed or removed?)"));
let rest = &src[start..];
let end = [
"\npub async fn ",
"\npub(crate) async fn ",
"\nasync fn ",
"\npub fn ",
"\nfn ",
]
.iter()
.filter_map(|p| rest[1..].find(p).map(|i| i + 1))
.min()
.unwrap_or(rest.len());
rest[..end]
.lines()
.filter(|l| !l.trim_start().starts_with("//"))
.collect::<Vec<_>>()
.join("\n")
}
#[test]
fn every_in_scope_mutation_has_its_gate() {
let pulls = include_str!("pulls.rs");
let webhooks = include_str!("webhooks.rs");
let labels = include_str!("labels.rs");
let issues = include_str!("issues.rs");
let bounties = include_str!("bounties.rs");
let replicas = include_str!("replicas.rs");
let tasks = include_str!("tasks.rs");
let stars = include_str!("stars.rs");
let protect = include_str!("protect.rs");
let visibility = include_str!("visibility.rs");
let profiles = include_str!("profiles.rs");
let repos = include_str!("repos.rs");
let register = include_str!("register.rs");
let ipfs = include_str!("ipfs.rs");
// (source, handler, expected gate marker)
let rows: &[(&str, &str, &str)] = &[
// Bucket A — owner-gate (require_repo_owner -> 403)
(pulls, "merge_pr", "require_repo_owner("),
(webhooks, "create_webhook", "require_repo_owner("),
(webhooks, "delete_webhook", "require_repo_owner("),
(labels, "add_label", "require_repo_owner("),
(labels, "remove_label", "require_repo_owner("),
// Bucket A' — owner OR author (did_matches against the author)
(pulls, "close_pr", "did_matches("),
(issues, "close_issue", "did_matches("),
// Bucket B — read-gate (authorize_repo_read)
(pulls, "create_review", "authorize_repo_read("),
(pulls, "create_comment", "authorize_repo_read("),
(pulls, "create_pr", "authorize_repo_read("),
(issues, "create_issue_comment", "authorize_repo_read("),
(issues, "create_issue", "authorize_repo_read("),
(bounties, "create_bounty", "authorize_repo_read("),
(repos, "fork_repo", "authorize_repo_read("),
// get_by_cid gates each iterated repo row directly via visibility_check
// (KTD2a: it must NOT route through authorize_repo_read's fuzzy re-resolve).
(ipfs, "get_by_cid", "visibility_check("),
// Bucket C — signer-self: the acting DID is matched/bound to auth.0
(tasks, "create_task", "did_matches("),
(tasks, "claim_task", "did_matches("),
(tasks, "complete_task", "did_matches("),
(tasks, "fail_task", "did_matches("),
(repos, "create_repo", "let owner_did = auth.0"),
(profiles, "set_profile", "let did = auth.0"),
(register, "register", "did_matches("),
(stars, "star_repo", "caller = &auth.0"),
(stars, "unstar_repo", "caller = &auth.0"),
// Bucket D — non-owner-by-design, positive per-route marker
(bounties, "claim_bounty", "claim_bounty(&id, &auth.0"),
(bounties, "submit_bounty", "did_matches("),
(bounties, "approve_bounty", "did_matches("),
(bounties, "cancel_bounty", "did_matches("),
(bounties, "dispute_bounty", "did_matches("),
(replicas, "register_replica", "did_matches("),
(replicas, "unregister_replica", "replica_did = &auth.0"),
// PRE-GATED — already owner-gated, in-scope group; guard the gate itself
(protect, "protect_branch", "did_matches("),
(protect, "unprotect_branch", "did_matches("),
(visibility, "set_visibility", "require_owner("),
(visibility, "remove_visibility", "require_owner("),
(visibility, "list_visibility", "require_owner("),
];
// The visibility rows prove require_owner is CALLED; this proves the helper
// itself does DID-safe matching, not a raw/trailing-segment compare.
assert!(
fn_body(visibility, "require_owner").contains("did_matches("),
"visibility::require_owner must use did_matches for DID-safe owner matching"
);
for (src, func, marker) in rows {
let body = fn_body(src, func);
assert!(
body.contains(marker),
"handler `{func}` is missing its gate marker `{marker}` — gate removed or route reclassified"
);
}
}
/// Proves the comment-stripping that GUARD-1 added: a marker that appears only
/// in a full-line comment (the real `replicas.rs` false-pass shape) must NOT
/// satisfy a row.
#[test]
fn comment_only_marker_does_not_satisfy_a_row() {
let src = "pub async fn demo() {\n // did_matches( handles the owner form\n do_thing();\n}\n";
assert!(
!fn_body(src, "demo").contains("did_matches("),
"a marker present only in a comment must not count as an enforced gate"
);
}
/// All `pub async fn` / `pub(crate) async fn` handler names declared in `src`.
/// Verb-agnostic on purpose: a repo-scoped read of any name is in scope, so a
/// handler named `fetch_*` / `replicate_*` / `info_refs` cannot escape the gate
/// check by not being called `list_*` / `get_*`.
fn handler_names(src: &str) -> Vec<String> {
["pub async fn ", "pub(crate) async fn "]
.iter()
.flat_map(|decl| {
src.match_indices(decl).map(move |(i, _)| {
src[i + decl.len()..]
.chars()
.take_while(|c| c.is_alphanumeric() || *c == '_')
.collect::<String>()
})
})
.collect()
}
/// True when the handler's signature takes an owner+repo path
/// (`Path<(String, String...)>`), i.e. it is a repo-scoped read rather than a
/// node-wide aggregate (`/stats`, `/ipfs/pins`, the global feeds).
fn is_repo_scoped(body: &str) -> bool {
let header = &body[..body.find('{').unwrap_or(body.len())];
header.contains("Path<(String, String")
}
/// True when at least one gate marker runs for EVERY service — i.e. some
/// marker sits outside any `if service == ...` discriminator block. A gate
/// that appears ONLY inside such a block (as the info/refs advertisement gate
/// did before #119: `visibility_check` ran under `if service ==
/// "git-upload-pack"`, leaving `git-receive-pack` ungated) covers a subset of
/// services and must NOT count as a full gate. Other handlers carry no
/// `service ==` discriminator, so for them this matches the plain
/// substring check. NOTE: only `if service ==` is detected — a
/// `match service { .. }` discriminator is NOT tracked and a gate inside one
/// arm would pass as full; avoid that shape, or extend the span loop below.
fn gate_runs_unconditionally(body: &str, markers: &[&str]) -> bool {
// Brace-matched spans of each `if service == ...` block.
let mut cond_spans: Vec<(usize, usize)> = Vec::new();
let mut search = 0;
while let Some(rel) = body[search..].find("if service ==") {
let cond_start = search + rel;
let Some(brace_rel) = body[cond_start..].find('{') else {
break;
};
let open = cond_start + brace_rel;
let mut depth = 0i32;
let mut end = body.len();
for (i, c) in body[open..].char_indices() {
match c {
'{' => depth += 1,
'}' => {
depth -= 1;
if depth == 0 {
end = open + i;
break;
}
}
_ => {}
}
}
cond_spans.push((open, end));
// On an unclosed block `end` stays at body.len() (the fail-safe
// direction: treat the rest as conditional rather than mask a gate);
// clamp so the next slice can't index past the end and panic.
search = (end + 1).min(body.len());
}
markers.iter().any(|m| {
body.match_indices(m)
.any(|(pos, _)| !cond_spans.iter().any(|(s, e)| pos >= *s && pos <= *e))
})
}
/// Egress gate guard: every repo-scoped handler (`Path<(String, String..)>`)
/// must carry an authz marker — a read gate (`authorize_repo_read` /
/// `visibility_check`), or a write gate (`require_repo_owner` / `require_owner`
/// / `did_matches` / a `&auth.0` self-binding) — or be listed in KNOWN_UNGATED
/// (the tracked, ungated reads). A new ungated handler of ANY verb trips this,
/// and a completeness scan over `src/api/` trips it for a whole new module that
/// adds a repo-scoped handler without being wired into `sources`. Mutations are
/// also checked precisely, per handler, by the mutation guard above; here they
/// only need *some* binding so the net stays verb-agnostic.
///
/// Scope and limits (this is a source scrape; the runtime route guard covers
/// behaviour):
/// - It proves a gate is CALLED, not that it runs on the requested path: a
/// content handler gating `"/"` instead of the subtree still passes here.
/// - It sees handlers taking the owner+repo tuple `Path<(String, String..)>`; a
/// repo identified by a struct/custom extractor would be missed (the helper
/// unit tests pin these boundaries).
/// - KNOWN_UNGATED entries need a real open issue and review; the staleness
/// assert removes one only once its gate lands.
#[test]
fn every_repo_scoped_handler_is_gated() {
let sources: &[(&str, &str)] = &[
(include_str!("bounties.rs"), "bounties.rs"),
(include_str!("certs.rs"), "certs.rs"),
(include_str!("changelog.rs"), "changelog.rs"),
(include_str!("encrypted.rs"), "encrypted.rs"),
(include_str!("events.rs"), "events.rs"),
(include_str!("issues.rs"), "issues.rs"),
(include_str!("labels.rs"), "labels.rs"),
(include_str!("protect.rs"), "protect.rs"),
(include_str!("pulls.rs"), "pulls.rs"),
(include_str!("replicas.rs"), "replicas.rs"),
(include_str!("repos.rs"), "repos.rs"),
(include_str!("stars.rs"), "stars.rs"),
(include_str!("visibility.rs"), "visibility.rs"),
(include_str!("webhooks.rs"), "webhooks.rs"),
];
let listed: std::collections::HashSet<&str> = sources.iter().map(|(_, f)| *f).collect();
assert!(!listed.is_empty(), "read-guard `sources` is empty");
// Completeness: every api/*.rs NOT already in `sources` must declare no
// repo-scoped handler, so a brand-new module cannot add an ungated handler
// the scrape never looks at. Reads the directory at test time (so the list
// cannot silently drift from the filesystem) and only inspects unlisted
// files — listed files are covered by the per-handler loop below.
let api_dir = concat!(env!("CARGO_MANIFEST_DIR"), "/src/api");
for (_, f) in sources {
assert!(
std::path::Path::new(api_dir).join(f).exists(),
"read-guard `sources` lists {f} but the file does not exist"
);
}
for entry in std::fs::read_dir(api_dir).expect("read src/api") {
let path = entry.expect("dir entry").path();
let fname = path.file_name().unwrap().to_string_lossy().into_owned();
if !fname.ends_with(".rs") || fname == "mod.rs" || listed.contains(fname.as_str()) {
continue;
}
let src = std::fs::read_to_string(&path).expect("read api file");
let has_repo_handler = handler_names(&src)
.iter()
.any(|n| is_repo_scoped(&fn_body(&src, n)));
assert!(
!has_repo_handler,
"api/{fname} declares a repo-scoped handler but is not in the egress \
guard `sources` list — add it so its handlers are gate-checked"
);
}
// Repo-scoped reads known to be ungated today, each tracked by an issue.
// Remove an entry the moment its gate lands (the staleness assert enforces it).
let known_ungated: &[(&str, &str)] = &[
// info/refs gates only git-upload-pack today; git-receive-pack
// advertisement is ungated until #119 makes the gate unconditional.
("git_info_refs", "#119"),
("list_certs", "#120"),
("get_cert", "#120"),
("list_issues", "#120"),
("get_issue", "#120"),
("list_issue_comments", "#120"),
("list_labels", "#120"),
("list_repo_bounties", "#120"),
("get_star_status", "#120"),
("list_repo_events", "#94 (PR #113)"),
("list_webhooks", "#94 (PR #113)"),
("list_replicas", "PR #113"),
("list_protected_branches", "PR #113"),
];
let is_known = |n: &str| known_ungated.iter().any(|(k, _)| *k == n);
// Any one of these = the handler binds the caller to an authz decision: the
// first two are read gates, the rest are the write/owner/self-binding forms.
// A repo-scoped mutation passes here on its write gate; the mutation guard
// above additionally verifies the exact gate type per handler. NOTE: a local
// rename of `caller`/`replica_did` must be paired with a marker update here.
let markers = [
"authorize_repo_read(",
"visibility_check(",
"require_repo_owner(",
"require_owner(",
"did_matches(",
"caller = &auth.0",
"replica_did = &auth.0",
];
// Every KNOWN_UNGATED name must be a real handler (catch typos / renames).
let all: std::collections::HashSet<String> =
sources.iter().flat_map(|(s, _)| handler_names(s)).collect();
for (n, _) in known_ungated {
assert!(
all.contains(*n),
"KNOWN_UNGATED lists `{n}`, which is not a real handler (renamed or removed?)"
);
}
let mut checked = 0usize;
for (src, file) in sources {
for name in handler_names(src) {
let body = fn_body(src, &name);
if !is_repo_scoped(&body) {
continue; // node-wide aggregate, not a repo-scoped surface
}
checked += 1;
let gated = gate_runs_unconditionally(&body, &markers);
assert!(
gated || is_known(&name),
"repo-scoped handler `{name}` ({file}) has no authz gate and is \
not in KNOWN_UNGATED — add the visibility/owner gate with the \
caller, or track it there"
);
assert!(
!(gated && is_known(&name)),
"handler `{name}` ({file}) is now gated — remove it from \
KNOWN_UNGATED so the allowlist stays an accurate gap list"
);
}
}
// Tripwire: if the scrape silently stopped finding handlers (e.g. a parser
// regression in handler_names/is_repo_scoped), this floor fails loudly
// instead of the guard passing vacuously. Current count is ~54 repo-scoped
// handlers; 20 is a deliberate floor that trips only on a gross regression.
assert!(
checked >= 20,
"egress guard only checked {checked} repo-scoped handlers — the scrape likely broke"
);
}
/// Pins the `handler_names` boundary: it collects every `pub`/`pub(crate)
/// async fn` regardless of verb (so a `fetch_*` read cannot escape), and skips
/// private `async fn` and sync `fn` helpers.
#[test]
fn handler_names_finds_all_pub_async_and_skips_others() {
let src = "pub async fn list_things() {}\n\
pub async fn fetch_thing() {}\n\
pub(crate) async fn get_crate_thing() {}\n\
async fn private_helper() {}\n\
fn sync_helper() {}\n";
let names = handler_names(src);
// Verb-agnostic: a non-list/get read verb is still seen.
assert!(names.contains(&"list_things".to_string()));
assert!(names.contains(&"fetch_thing".to_string()));
// pub(crate) routed handlers are in scope too.
assert!(names.contains(&"get_crate_thing".to_string()));
// Private/sync helpers are not routed handlers and are skipped.
assert!(!names
.iter()
.any(|n| n == "private_helper" || n == "sync_helper"));
}
/// Pins the `is_repo_scoped` boundary: an owner+repo tuple Path is repo-scoped;
/// a single-segment or absent Path is a node-wide aggregate.
#[test]
fn is_repo_scoped_requires_owner_repo_path() {
let repo =
"fn get_x(s: State, Path((owner, name)): Path<(String, String)>) {\n body();\n}";
let three = "fn get_y(Path((o, n, id)): Path<(String, String, String)>) {\n body();\n}";
let node_wide = "fn list_z(s: State<AppState>) {\n body();\n}";
let single = "fn get_w(Path(cid): Path<String>) {\n body();\n}";
assert!(is_repo_scoped(repo), "owner+repo tuple is repo-scoped");
assert!(is_repo_scoped(three), "owner+repo+id tuple is repo-scoped");
assert!(!is_repo_scoped(node_wide), "no Path is node-wide");
assert!(
!is_repo_scoped(single),
"single-segment Path is not repo-scoped"
);
}
/// Pins `gate_runs_unconditionally`: a gate nested only inside an
/// `if service == ...` block is conditional (does NOT count), while the same
/// gate at the top level — or an additional unconditional one — does.
#[test]
fn conditional_service_gate_is_not_a_full_gate() {
let markers = ["visibility_check("];
// Gate runs only for one service: not a full gate.
let conditional = "fn f() {\n \
if service == \"git-upload-pack\" {\n \
visibility_check(rules, caller);\n \
}\n \
if service == \"git-receive-pack\" { acquire_fresh(); }\n}";
assert!(
!gate_runs_unconditionally(conditional, &markers),
"a gate only inside `if service ==` covers a subset of services"
);
// Same gate at top level: full gate.
let unconditional = "fn f() {\n \
visibility_check(rules, caller);\n \
if service == \"git-receive-pack\" { acquire_fresh(); }\n}";
assert!(
gate_runs_unconditionally(unconditional, &markers),
"an unconditional gate runs for every service"
);
// A gate inside EACH of two service blocks, none outside: still a
// subset (no service clears it unconditionally), so not a full gate.
let both_conditional = "fn f() {\n \
if service == \"git-upload-pack\" { visibility_check(a); }\n \
if service == \"git-receive-pack\" { visibility_check(b); }\n}";
assert!(
!gate_runs_unconditionally(both_conditional, &markers),
"a gate inside every service block is still conditional"
);
// A marker inside one block AND again unconditionally: the
// unconditional occurrence makes it a full gate (exercises the
// match_indices scan past the in-block hit).
let inside_and_outside = "fn f() {\n \
if service == \"git-upload-pack\" { visibility_check(a); }\n \
visibility_check(b);\n}";
assert!(
gate_runs_unconditionally(inside_and_outside, &markers),
"an unconditional occurrence counts even when another is conditional"
);
// No marker at all: not gated.
assert!(!gate_runs_unconditionally(
"fn f() { do_thing(); }",
&markers
));
// An unclosed `if service ==` block (e.g. phantom brace from a string
// literal) must not panic on the slice advance; the span runs to EOF, so
// the in-block marker reads as conditional. Real Rust source is balanced,
// so this only guards the scraper against a future pathological body.
let unclosed = "fn f() { if service == \"x\" { visibility_check(a);";
assert!(
!gate_runs_unconditionally(unclosed, &markers),
"an unclosed service block must not panic and stays conditional"
);
}
}
+218 -4
View File
@@ -40,6 +40,123 @@ pub struct PeerResponse {
pub reachable: bool,
}
/// Extract an IPv4 address embedded in an IPv6 literal across the transition
/// formats that carry one: IPv4-mapped (`::ffff:a.b.c.d`), IPv4-compatible
/// (`::a.b.c.d`), 6to4 (`2002:WWXX:YYZZ::/16`), and the NAT64 well-known prefix
/// (`64:ff9b::/96`). Returns `None` for native IPv6. Callers fold the result
/// back through the v4 range checks so loopback/private addresses smuggled in
/// via any of these encodings are rejected, not just the mapped/compatible pair.
///
/// INCOMPLETE-NAT64 CONTRACT: non-well-known NAT64 prefixes (e.g. the RFC 8215
/// local-use `64:ff9b:1::/48`) are deliberately **not** decoded here — their v4
/// sits at a prefix-length-dependent offset (RFC 6052 §2.2) — so they return
/// `None`. Any caller that needs them blocked must reject the wider
/// `64:ff9b::/32` itself; `is_public_http_url` does this in its native-v6 arm.
fn embedded_ipv4(v6: std::net::Ipv6Addr) -> Option<std::net::Ipv4Addr> {
if let Some(v4) = v6.to_ipv4_mapped().or_else(|| v6.to_ipv4()) {
return Some(v4);
}
let s = v6.segments();
// 6to4 (2002::/16) embeds W.X.Y.Z in the next two segments.
if s[0] == 0x2002 {
let [a, b] = s[1].to_be_bytes();
let [c, d] = s[2].to_be_bytes();
return Some(std::net::Ipv4Addr::new(a, b, c, d));
}
// NAT64 well-known prefix (64:ff9b::/96) embeds the IPv4 in the low 32 bits.
if s[0] == 0x0064 && s[1] == 0xff9b && s[2..6] == [0, 0, 0, 0] {
let [a, b] = s[6].to_be_bytes();
let [c, d] = s[7].to_be_bytes();
return Some(std::net::Ipv4Addr::new(a, b, c, d));
}
None
}
/// Whether a peer `http_url` is a public http(s) endpoint safe to register.
/// Rejects non-http(s) schemes, loopback/unspecified/private/link-local IPs,
/// and `localhost` / `.local` / `.internal` hostnames. Used at announce time
/// and by the boot-time prune of already-poisoned rows.
pub fn is_public_http_url(raw: &str) -> bool {
let url = match reqwest::Url::parse(raw) {
Ok(u) => u,
Err(_) => return false,
};
if !matches!(url.scheme(), "http" | "https") {
return false;
}
let mut host = match url.host_str() {
Some(h) => h.to_ascii_lowercase(),
None => return false,
};
// Drop a single trailing dot (FQDN root): `localhost.` resolves the same as
// `localhost`, so normalize before the suffix/equality checks below.
if let Some(stripped) = host.strip_suffix('.') {
host = stripped.to_string();
}
if host.is_empty()
|| host == "localhost"
|| host.ends_with(".local")
|| host.ends_with(".internal")
{
return false;
}
// host_str() keeps brackets on IPv6 literals (e.g. "[::1]"); strip them
// before parsing as an IP.
let ip_candidate = host.trim_start_matches('[').trim_end_matches(']');
if let Ok(ip) = ip_candidate.parse::<std::net::IpAddr>() {
// Reject loopback/unspecified on the literal as given — catches `::1`
// and `::` before the IPv4-folding below (`::1`.to_ipv4() would
// otherwise map to a non-loopback `0.0.0.1`).
if ip.is_loopback() || ip.is_unspecified() {
return false;
}
// Fold any IPv6 literal that embeds an IPv4 address (mapped, compatible,
// 6to4, NAT64) down to that IPv4 so the v4 range checks catch
// loopback/private addresses smuggled in via an IPv6 encoding, then
// re-check loopback/unspecified.
let ip = match ip {
std::net::IpAddr::V6(v6) => embedded_ipv4(v6).map(std::net::IpAddr::V4).unwrap_or(ip),
v4 => v4,
};
if ip.is_loopback() || ip.is_unspecified() {
return false;
}
match ip {
std::net::IpAddr::V4(v4) => {
let o = v4.octets();
// RFC1918 private, link-local, CGNAT (100.64.0.0/10), or the
// RFC1122 "this host" block 0.0.0.0/8 (never a valid destination;
// 0.0.0.0 itself is already caught by the is_unspecified check).
if v4.is_private()
|| v4.is_link_local()
|| (o[0] == 100 && (o[1] & 0xc0) == 64)
|| o[0] == 0
{
return false;
}
}
std::net::IpAddr::V6(v6) => {
let s = v6.segments();
// fc00::/7 (unique-local) or fe80::/10 (link-local)
if (s[0] & 0xfe00) == 0xfc00 || (s[0] & 0xffc0) == 0xfe80 {
return false;
}
// Any NAT64 address (64:ff9b::/32) that is not the cleanly
// decodable well-known /96 — e.g. the RFC 8215 local-use
// 64:ff9b:1::/48 — carries a translated target whose embedded v4
// sits at a prefix-length-dependent offset (RFC 6052 §2.2).
// Rather than risk a wrong decode across every prefix length we
// reject the whole NAT64 space here. The well-known /96 was
// already folded to its v4 above and never reaches this arm.
if s[0] == 0x0064 && s[1] == 0xff9b {
return false;
}
}
}
}
true
}
/// GET /api/v1/peers
pub async fn list_peers(State(state): State<AppState>) -> Result<Json<serde_json::Value>> {
let peers = state.db.list_peers().await?;
@@ -83,10 +200,14 @@ pub async fn announce(
);
}
// Validate the URL is HTTP/HTTPS
if !req.http_url.starts_with("http://") && !req.http_url.starts_with("https://") {
// Validate the URL is a public http(s) endpoint. The announce route is
// reachable unauthenticated (until all peers sign), so without this an
// attacker can register loopback/private "peers" (localhost:5432, etc.)
// and turn our outbound sync-notify fan-out into an SSRF probe — and bury
// the real peers under junk so node-origin repos stop replicating.
if !is_public_http_url(&req.http_url) {
return Err(AppError::BadRequest(
"http_url must start with http:// or https://".into(),
"http_url must be a public http(s) URL (no loopback, private, or .internal/.local hosts)".into(),
));
}
@@ -311,7 +432,13 @@ pub async fn ping_peer(
// Async ping
let url = format!("{}/health", peer.http_url.trim_end_matches('/'));
let ok = reqwest::get(&url)
// Use the shared no-redirect client: bare `reqwest::get` follows redirects,
// so a peer could answer with `302 -> http://127.0.0.1/` and turn the ping
// into an SSRF probe.
let ok = state
.http_client
.get(&url)
.send()
.await
.map(|r| r.status().is_success())
.unwrap_or(false);
@@ -324,3 +451,90 @@ pub async fn ping_peer(
"reachable": ok,
})))
}
#[cfg(test)]
mod tests {
use super::is_public_http_url;
#[test]
fn accepts_public_https_and_http() {
assert!(is_public_http_url("https://node.gitlawb.com"));
assert!(is_public_http_url("https://manila.gitlawb.com/"));
assert!(is_public_http_url("http://203.0.113.10:7545"));
}
#[test]
fn rejects_loopback_private_and_internal() {
for bad in [
"http://localhost:7545",
"http://127.0.0.1:5432/",
"http://localhost:22/",
"http://0.0.0.0:7545",
// RFC1122 "this host" block 0.0.0.0/8 beyond 0.0.0.0 itself,
// including the upper boundary (1.0.0.0 is public, tested below).
"http://0.0.0.1:7545",
"http://0.1.2.3/",
"http://0.255.255.255/",
"http://10.0.0.5:7545",
"http://192.168.1.10/",
"http://172.16.0.1:7545",
"http://169.254.1.1/",
"http://[::1]:7545",
"http://gitlawb-node.internal:7545",
"http://my-node.local/",
"ftp://node.gitlawb.com",
"not-a-url",
"",
// Trailing-dot FQDN of an internal host.
"http://localhost./",
// CGNAT (100.64.0.0/10).
"http://100.64.0.1:7545",
"http://100.127.255.255/",
// IPv4-mapped / IPv4-compatible IPv6 smuggling private/loopback v4.
"http://[::ffff:127.0.0.1]:7545",
"http://[::ffff:10.0.0.1]/",
"http://[::ffff:192.168.1.1]:7545",
"http://[::127.0.0.1]/",
// 6to4 (2002::/16) embedding loopback/private v4.
"http://[2002:7f00:1::]:7545",
"http://[2002:a00:1::]/",
"http://[2002:c0a8:101::]/",
// NAT64 well-known prefix (64:ff9b::/96) embedding loopback/private v4.
"http://[64:ff9b::7f00:1]/",
"http://[64:ff9b::a00:1]:7545",
// NAT64 local-use prefix (64:ff9b:1::/48, RFC 8215) — rejected
// outright rather than decoded, so the whole NAT64 space is closed.
"http://[64:ff9b:1::7f00:1]/",
"http://[64:ff9b:1::a00:1]:7545",
"http://[64:ff9b:1:a00::]/",
// Local-use NAT64 wrapping a *public* v4 (203.0.113.10) is still
// rejected: the block is prefix-based, not payload-based.
"http://[64:ff9b:1::cb00:710a]/",
] {
assert!(!is_public_http_url(bad), "{bad:?} must be rejected");
}
}
#[test]
fn accepts_6to4_and_nat64_wrapping_public_v4() {
// Fold-and-recheck stays consistent with the mapped/compatible handling:
// a transition address that wraps a public v4 is allowed, only the
// private/loopback-wrapping forms above are rejected. 203.0.113.10.
assert!(is_public_http_url("http://[2002:cb00:710a::]/"));
assert!(is_public_http_url("http://[64:ff9b::cb00:710a]/"));
}
#[test]
fn accepts_public_outside_cgnat_range() {
// 100.0.0.0/10 boundary: .63 and .128 are public, only 100.64-127 is CGNAT.
assert!(is_public_http_url("http://100.63.255.255:7545"));
assert!(is_public_http_url("http://100.128.0.1/"));
}
#[test]
fn accepts_first_address_above_this_host_block() {
// 1.0.0.0 is the first address above the rejected 0.0.0.0/8 block;
// pins the `o[0] == 0` check against an off-by-one into 1.x.
assert!(is_public_http_url("http://1.0.0.0/"));
}
}
+5 -15
View File
@@ -23,15 +23,10 @@ pub async fn protect_branch(
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{repo}")))?;
// Only the repo owner can protect branches
// Only the repo owner can protect branches (DID-safe match, shared idiom).
let caller = &auth.0;
let owner_short = record
.owner_did
.split(':')
.next_back()
.unwrap_or(&record.owner_did);
if caller != &record.owner_did && caller != owner_short {
return Err(AppError::BadRequest(
if !crate::api::did_matches(caller, &record.owner_did) {
return Err(AppError::Forbidden(
"only the repo owner can protect branches".into(),
));
}
@@ -63,13 +58,8 @@ pub async fn unprotect_branch(
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{repo}")))?;
let caller = &auth.0;
let owner_short = record
.owner_did
.split(':')
.next_back()
.unwrap_or(&record.owner_did);
if caller != &record.owner_did && caller != owner_short {
return Err(AppError::BadRequest(
if !crate::api::did_matches(caller, &record.owner_did) {
return Err(AppError::Forbidden(
"only the repo owner can unprotect branches".into(),
));
}
+26 -15
View File
@@ -40,11 +40,11 @@ pub async fn create_pr(
Path((owner, name)): Path<(String, String)>,
Json(req): Json<CreatePrRequest>,
) -> Result<(StatusCode, Json<PullRequest>)> {
let record = state
.db
.get_repo(&owner, &name)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
// Authorize the caller as a reader before accepting a PR: a non-reader must
// not be able to open a PR (and fire its webhooks) against a private repo
// they cannot read. Mirrors create_review / create_comment / create_bounty.
let (record, _rules) =
crate::api::authorize_repo_read(&state, &owner, &name, Some(auth.0.as_str()), "/").await?;
let author_did = auth.0;
let target_branch = req
@@ -194,6 +194,11 @@ pub async fn merge_pr(
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
// Owner-only merge (N7). Merging writes the served tree, so this is the same
// trust boundary as owner-only push; it subsumes branch protection (a
// non-owner cannot merge to any branch, protected or not).
crate::api::require_repo_owner(&record, &auth.0)?;
let pr = state
.db
.get_pr(&record.id, number)
@@ -252,6 +257,7 @@ pub async fn merge_pr(
/// POST /api/v1/repos/:owner/:repo/pulls/:number/close
pub async fn close_pr(
State(state): State<AppState>,
Extension(auth): Extension<AuthenticatedDid>,
Path((owner, name, number)): Path<(String, String, i64)>,
) -> Result<Json<serde_json::Value>> {
let record = state
@@ -266,6 +272,15 @@ pub async fn close_pr(
.await?
.ok_or_else(|| AppError::NotFound(format!("PR #{number} not found")))?;
// Owner OR author may close (forge norm).
let is_owner = crate::api::require_repo_owner(&record, &auth.0).is_ok();
let is_author = crate::api::did_matches(&auth.0, &pr.author_did);
if !is_owner && !is_author {
return Err(AppError::Forbidden(
"only the repo owner or the PR author can close this PR".into(),
));
}
state.db.close_pr(&pr.id).await?;
webhooks::fire_event(
@@ -290,11 +305,9 @@ pub async fn create_review(
Path((owner, name, number)): Path<(String, String, i64)>,
Json(req): Json<CreateReviewRequest>,
) -> Result<(StatusCode, Json<PrReview>)> {
let record = state
.db
.get_repo(&owner, &name)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
// Read-gate: a reviewer must be able to read the repo, but need not own it.
let (record, _rules) =
crate::api::authorize_repo_read(&state, &owner, &name, Some(auth.0.as_str()), "/").await?;
let pr = state
.db
@@ -369,11 +382,9 @@ pub async fn create_comment(
));
}
let record = state
.db
.get_repo(&owner, &name)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
// Read-gate: a commenter must be able to read the repo, but need not own it.
let (record, _rules) =
crate::api::authorize_repo_read(&state, &owner, &name, Some(auth.0.as_str()), "/").await?;
let pr = state
.db
+18
View File
@@ -35,14 +35,32 @@ pub struct RegisterResponse {
/// Accepts only requests with a valid HTTP Signature (enforced by middleware).
pub async fn register(
State(state): State<AppState>,
axum::Extension(auth): axum::Extension<crate::auth::AuthenticatedDid>,
headers: axum::http::HeaderMap,
Json(req): Json<RegisterRequest>,
) -> Result<(StatusCode, Json<RegisterResponse>)> {
// iCaptcha gate (inert unless ICAPTCHA_MODE is set). Verify up front; spend
// the proof only once the request is admissible, just before the write, so a
// rejected request never burns a valid proof.
let proof = crate::icaptcha::verify_request(&headers, &auth.0)?;
// Parse and validate the DID
let agent_did: Did = req
.did
.parse()
.map_err(|e: gitlawb_core::Error| AppError::BadRequest(e.to_string()))?;
// Bind registration to the authenticated signer: an agent may only register
// itself, not create or refresh a trust row for a DID it does not control.
if !crate::api::did_matches(&auth.0, agent_did.as_str()) {
return Err(AppError::Forbidden(
"did must be the authenticated signer".into(),
));
}
// Request is admissible — spend the proof now, immediately before the write.
proof.consume(&state.db).await?;
// Store the agent in the local index
state
.db
+3 -2
View File
@@ -47,8 +47,9 @@ pub async fn register_replica(
let replica_did = &auth.0;
// Don't let an owner register themselves as a replica of their own repo.
if replica_did == &record.owner_did {
// Don't let an owner register themselves as a replica of their own repo
// (did_matches handles the full vs bare did:key owner form).
if crate::api::did_matches(replica_did, &record.owner_did) {
return Err(AppError::BadRequest(
"the repo owner is not a replica of their own repo".into(),
));
File diff suppressed because it is too large. Load diff
+85 -19
View File
@@ -9,7 +9,7 @@
//! POST /api/v1/tasks/{id}/fail — fail task
use axum::{
extract::{Path, Query, State},
extract::{Extension, Path, Query, State},
http::StatusCode,
Json,
};
@@ -18,9 +18,18 @@ use serde::Deserialize;
use serde_json::{json, Value};
use uuid::Uuid;
use crate::auth::AuthenticatedDid;
use crate::db::AgentTask;
use crate::state::{AppState, TaskEventBroadcast};
/// 403 in this module's error shape (`(StatusCode, Json<Value>)`, not `AppError`).
fn forbidden(msg: &str) -> (StatusCode, Json<Value>) {
(
StatusCode::FORBIDDEN,
Json(json!({ "error": "forbidden", "message": msg })),
)
}
// ── Request / response types ──────────────────────────────────────────────────
#[derive(Deserialize)]
@@ -55,13 +64,11 @@ pub struct ClaimTaskBody {
#[derive(Deserialize)]
pub struct CompleteTaskBody {
pub result: Option<String>,
pub by_did: Option<String>,
}
#[derive(Deserialize)]
pub struct FailTaskBody {
pub reason: Option<String>,
pub by_did: Option<String>,
}
fn task_to_json(t: &AgentTask) -> Value {
@@ -87,15 +94,20 @@ fn task_to_json(t: &AgentTask) -> Value {
/// POST /api/v1/tasks
pub async fn create_task(
State(state): State<AppState>,
Extension(auth): Extension<AuthenticatedDid>,
Json(body): Json<CreateTaskBody>,
) -> Result<(StatusCode, Json<Value>), (StatusCode, Json<Value>)> {
// Bind the delegator to the authenticated signer (N13).
if !crate::api::did_matches(&auth.0, &body.delegator_did) {
return Err(forbidden("delegator_did must be the authenticated signer"));
}
let now = Utc::now().to_rfc3339();
let task = AgentTask {
id: Uuid::new_v4().to_string(),
repo_id: body.repo_id,
kind: body.kind,
status: "pending".to_string(),
delegator_did: body.delegator_did,
delegator_did: auth.0,
assignee_did: body.assignee_did,
capability: body.capability,
ucan_token: body.ucan_token,
@@ -154,24 +166,25 @@ pub async fn get_task(
/// POST /api/v1/tasks/{id}/claim
pub async fn claim_task(
State(state): State<AppState>,
Extension(auth): Extension<AuthenticatedDid>,
Path(id): Path<String>,
Json(body): Json<ClaimTaskBody>,
) -> Result<Json<Value>, (StatusCode, Json<Value>)> {
let task = state
.db
.claim_task(&id, &body.assignee_did)
.await
.map_err(|e| {
(
StatusCode::CONFLICT,
Json(json!({ "error": e.to_string() })),
)
})?;
// Bind the assignee to the authenticated signer (N13).
if !crate::api::did_matches(&auth.0, &body.assignee_did) {
return Err(forbidden("assignee_did must be the authenticated signer"));
}
let task = state.db.claim_task(&id, &auth.0).await.map_err(|e| {
(
StatusCode::CONFLICT,
Json(json!({ "error": e.to_string() })),
)
})?;
let _ = state.task_event_tx.send(TaskEventBroadcast {
task_id: id,
old_status: "pending".to_string(),
new_status: "claimed".to_string(),
by_did: body.assignee_did,
by_did: auth.0,
at: Utc::now().to_rfc3339(),
});
Ok(Json(task_to_json(&task)))
@@ -180,17 +193,44 @@ pub async fn claim_task(
/// POST /api/v1/tasks/{id}/complete
pub async fn complete_task(
State(state): State<AppState>,
Extension(auth): Extension<AuthenticatedDid>,
Path(id): Path<String>,
Json(body): Json<CompleteTaskBody>,
) -> Result<Json<Value>, (StatusCode, Json<Value>)> {
let by_did = body.by_did.unwrap_or_default();
// Authorize the actor, not just bind their identity: the N13 signer-binding
// proved the caller was whoever they claimed, but never that they were the
// task's assignee. Load the task and require the caller to be its assignee;
// finish_task then transitions only a claimed task.
let existing = state
.db
.get_task(&id)
.await
.map_err(|e| {
(
StatusCode::INTERNAL_SERVER_ERROR,
Json(json!({ "error": e.to_string() })),
)
})?
.ok_or_else(|| {
(
StatusCode::NOT_FOUND,
Json(json!({ "error": "task not found" })),
)
})?;
if !crate::api::did_matches(
&auth.0,
existing.assignee_did.as_deref().unwrap_or_default(),
) {
return Err(forbidden("only the task assignee can complete it"));
}
let by_did = auth.0;
let task = state
.db
.finish_task(&id, "completed", body.result.as_deref())
.await
.map_err(|e| {
(
StatusCode::UNPROCESSABLE_ENTITY,
StatusCode::CONFLICT,
Json(json!({ "error": e.to_string() })),
)
})?;
@@ -207,10 +247,36 @@ pub async fn complete_task(
/// POST /api/v1/tasks/{id}/fail
pub async fn fail_task(
State(state): State<AppState>,
Extension(auth): Extension<AuthenticatedDid>,
Path(id): Path<String>,
Json(body): Json<FailTaskBody>,
) -> Result<Json<Value>, (StatusCode, Json<Value>)> {
let by_did = body.by_did.unwrap_or_default();
// Authorize the actor, not just bind their identity (see complete_task): only
// the task's assignee may fail it, and finish_task transitions only a claimed
// task.
let existing = state
.db
.get_task(&id)
.await
.map_err(|e| {
(
StatusCode::INTERNAL_SERVER_ERROR,
Json(json!({ "error": e.to_string() })),
)
})?
.ok_or_else(|| {
(
StatusCode::NOT_FOUND,
Json(json!({ "error": "task not found" })),
)
})?;
if !crate::api::did_matches(
&auth.0,
existing.assignee_did.as_deref().unwrap_or_default(),
) {
return Err(forbidden("only the task assignee can fail it"));
}
let by_did = auth.0;
let reason = body.reason.unwrap_or_default();
let task = state
.db
@@ -218,7 +284,7 @@ pub async fn fail_task(
.await
.map_err(|e| {
(
StatusCode::UNPROCESSABLE_ENTITY,
StatusCode::CONFLICT,
Json(json!({ "error": e.to_string() })),
)
})?;
+40 -14
View File
@@ -26,17 +26,16 @@ pub struct RemoveVisibilityRequest {
}
fn require_owner(record: &crate::db::RepoRecord, caller: &str) -> Result<()> {
let owner_short = record
.owner_did
.split(':')
.next_back()
.unwrap_or(&record.owner_did);
if caller != record.owner_did && caller != owner_short {
return Err(AppError::BadRequest(
// DID-safe owner match (collapses did:key full vs bare on both sides, never
// across methods), shared with require_repo_owner — not a trailing-segment
// compare that only normalized the owner side.
if crate::api::did_matches(caller, &record.owner_did) {
Ok(())
} else {
Err(AppError::Forbidden(
"only the repo owner can manage visibility".into(),
));
))
}
Ok(())
}
/// Reject malformed globs before they reach the store, where they would
@@ -57,9 +56,23 @@ fn validate_path_glob(path_glob: &str) -> Result<()> {
"path_glob must not end with '/'".into(),
));
}
if path_glob.contains('*') && !path_glob.ends_with("/**") {
// Reject empty path segments. `glob_prefix` collapses "//**" to "/" (whole
// repo), so without this an operator could bypass the "/**" guard above and
// grant repo-wide read with "//**", or set a dead rule like "//secret/**"
// whose prefix never matches a real single-slash path and silently leaves
// the subtree readable.
if path_glob.contains("//") {
return Err(AppError::BadRequest(
"the only supported wildcard is a trailing '/**'".into(),
"path_glob must not contain empty path segments".into(),
));
}
// Strip the one permitted trailing "/**" before checking for stray
// wildcards, otherwise a "*" elsewhere in the prefix (e.g. "/a*b/**" or
// "/a/**/**") slips through because the string still ends with "/**".
let prefix = path_glob.strip_suffix("/**").unwrap_or(path_glob);
if prefix.contains('*') {
return Err(AppError::BadRequest(
"the only supported wildcard is a single trailing '/**'".into(),
));
}
Ok(())
@@ -240,9 +253,22 @@ mod tests {
#[test]
fn rejects_malformed_globs() {
// empty, no leading slash, whole-repo via "/**", trailing slash, and
// non-trailing wildcards are all rejected.
for g in ["", "secret/**", "/**", "/secret/", "/a*b", "/*/x"] {
// empty, no leading slash, whole-repo via "/**", trailing slash,
// non-trailing wildcards, double wildcards, and empty path segments
// (which would collapse to whole-repo or a dead rule) are all rejected.
for g in [
"",
"secret/**",
"/**",
"/**/**",
"/secret/",
"/a*b",
"/*/x",
"/a*b/**",
"/a/**/**",
"//**",
"//secret/**",
] {
assert!(validate_path_glob(g).is_err(), "{g} should be rejected");
}
}
+42 -3
View File
@@ -38,11 +38,16 @@ pub async fn create_webhook(
.get_repo(&owner, &name)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
crate::api::require_repo_owner(&record, &auth.0)?;
// Validate URL is http/https
if !req.url.starts_with("http://") && !req.url.starts_with("https://") {
// Gate the target through the same hardened public-host validator the peer
// announce path uses, so an authenticated owner cannot register a webhook
// that makes the node POST to loopback/private/link-local/metadata
// endpoints (SSRF). Delivery runs on the shared no-redirect client
// (main.rs), which closes the 3xx-to-internal bounce.
if !crate::api::peers::is_public_http_url(&req.url) {
return Err(AppError::BadRequest(
"webhook URL must be http:// or https://".into(),
"webhook URL must be a public http(s) URL (no loopback, private, or .internal/.local hosts)".into(),
));
}
@@ -95,6 +100,7 @@ pub async fn list_webhooks(
/// DELETE /api/v1/repos/:owner/:repo/hooks/:id
pub async fn delete_webhook(
State(state): State<AppState>,
Extension(auth): Extension<AuthenticatedDid>,
Path((owner, name, id)): Path<(String, String, String)>,
) -> Result<Json<serde_json::Value>> {
let record = state
@@ -102,6 +108,7 @@ pub async fn delete_webhook(
.get_repo(&owner, &name)
.await?
.ok_or_else(|| AppError::RepoNotFound(format!("{owner}/{name}")))?;
crate::api::require_repo_owner(&record, &auth.0)?;
// Verify the webhook belongs to this repo
let hook = state
@@ -117,3 +124,35 @@ pub async fn delete_webhook(
state.db.delete_webhook(&id).await?;
Ok(Json(serde_json::json!({ "deleted": true, "id": id })))
}
#[cfg(test)]
mod tests {
use crate::api::peers::is_public_http_url;
// create_webhook gates req.url through is_public_http_url. Pin the exact
// SSRF targets from issue #81 so the webhook path can never regress to the
// old scheme-only check, independent of the validator's own peer tests.
#[test]
fn webhook_url_gate_rejects_ssrf_targets() {
for bad in [
"http://127.0.0.1:5432/",
"http://169.254.169.254/latest/meta-data/",
"http://localhost/",
"http://10.0.0.5/",
"http://[::1]/",
// IPv6 transition encodings smuggling loopback v4 (6to4 / NAT64).
"http://[2002:7f00:1::]/",
"http://[64:ff9b::7f00:1]/",
"ftp://example.com/",
"not-a-url",
] {
assert!(!is_public_http_url(bad), "{bad:?} must be rejected");
}
}
#[test]
fn webhook_url_gate_allows_public_targets() {
assert!(is_public_http_url("https://hooks.example.com/gitlawb"));
assert!(is_public_http_url("http://203.0.113.10:7545/"));
}
}
+40 -2
View File
@@ -243,7 +243,7 @@ mod tests {
repo: "alice/myrepo".into(),
owner_did: "did:key:z6Mk...".into(),
ref_name: "refs/heads/main".into(),
old_sha: "0000000000000000000000000000000000000000000000000000000000000000".into(),
old_sha: "0000000000000000000000000000000000000000".into(),
new_sha: "a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2c3d4e5f6a1b2".into(),
cid: Some("bafyreib5...".into()),
timestamp: "2026-03-14T00:00:00Z".into(),
@@ -270,7 +270,7 @@ mod tests {
repo: "alice/myrepo".into(),
owner_did: "did:key:z6Mk...".into(),
ref_name: "refs/heads/main".into(),
old_sha: "0".repeat(64),
old_sha: "0".repeat(40),
new_sha: "a1b2c3d4".repeat(8),
cid: None,
timestamp: "2026-03-14T00:00:00Z".into(),
@@ -286,6 +286,44 @@ mod tests {
_mock.assert_async().await;
}
#[tokio::test]
async fn test_anchor_body_carries_real_old_sha() {
// The anchored body must serialize the real old→new transition the
// node was handed, never a zero placeholder. Regression guard for the
// push handler that used to hardcode `old_sha` to 64 zeros (#26).
let mut server = mockito::Server::new_async().await;
let real_old = "1111111111111111111111111111111111111111";
let real_new = "2222222222222222222222222222222222222222";
let _mock = server
.mock("POST", "/upload")
.match_body(mockito::Matcher::AllOf(vec![
mockito::Matcher::PartialJsonString(format!(r#"{{"old_sha":"{real_old}"}}"#)),
mockito::Matcher::PartialJsonString(format!(r#"{{"new_sha":"{real_new}"}}"#)),
]))
.with_status(200)
.with_header("content-type", "application/json")
.with_body(r#"{"id":"TX_REAL_OLD_SHA","timestamp":1710000000000,"version":"1.0.0"}"#)
.create_async()
.await;
let client = reqwest::Client::new();
let anchor = RefAnchor {
repo: "alice/myrepo".into(),
owner_did: "did:key:z6Mk...".into(),
ref_name: "refs/heads/main".into(),
old_sha: real_old.into(),
new_sha: real_new.into(),
cid: None,
timestamp: "2026-03-14T00:00:00Z".into(),
node_did: "did:key:z6Mknnd...".into(),
};
let result = anchor_ref_update(&client, &server.url(), &anchor).await;
assert_eq!(result.unwrap(), "TX_REAL_OLD_SHA");
// The mock only matches when the posted JSON carries both real SHAs.
_mock.assert_async().await;
}
#[test]
fn test_arweave_url() {
let url = arweave_url("7xGpIoHUQ8j9GhD3Y2mKzP1NsVtXwRcFe4bEaLnMuOk");
+12
View File
@@ -17,6 +17,18 @@ use crate::state::AppState;
#[derive(Clone, Debug)]
pub struct AuthenticatedDid(pub String);
/// Whether `caller` is authorized to push to `record`.
///
/// Phase 1 (`GITLAWB_ENFORCE_OWNER_PUSH`): owner-only, via the canonical
/// [`crate::api::did_matches`] owner comparison (DID-safe on both sides). This is
/// intentionally a distinct, intent-named gate rather than a bare owner check so
/// that Phase 2 can extend it to honor a verified UCAN `git/push` capability as a
/// pure addition (`did_matches(..) || ucan_grants_push(..)`) without rewriting
/// call sites.
pub fn caller_authorized_to_push(record: &crate::db::RepoRecord, caller: &str) -> bool {
crate::api::did_matches(caller, &record.owner_did)
}
use gitlawb_core::http_sig::{
build_signing_string, compute_content_digest, HttpSignature, COVERED_COMPONENTS,
};
+8
View File
@@ -49,6 +49,14 @@ pub struct Config {
)]
pub require_signed_peer_writes: bool,
/// Require the authenticated pusher to be the repo owner on `git-receive-pack`.
/// Authentication (a valid did:key signature) is not authorization on its own:
/// any party can sign as their own DID. When true, pushes whose authenticated
/// DID is not the repo owner are rejected. Keep false during rolling upgrades;
/// flip it on once owners are ready for owner-only writes.
#[arg(long, env = "GITLAWB_ENFORCE_OWNER_PUSH", default_value_t = false)]
pub enforce_owner_push: bool,
/// URL of local IPFS/Kubo node HTTP API (e.g. http://127.0.0.1:5001)
#[arg(long, env = "GITLAWB_IPFS_API", default_value = "")]
pub ipfs_api: String,
File diff suppressed because it is too large. Load diff
+250 -24
View File
@@ -38,6 +38,70 @@ fn did_to_key(did: &str) -> Option<VerifyingKey> {
Did::from_str(did).ok()?.to_verifying_key().ok()
}
/// Resolve every recipient DID to its verifying key, all-or-nothing.
///
/// Returns `Ok(keys)` only when every DID resolves. If any DID fails, returns
/// `Err(unresolved)` listing the unresolvable DID strings so the caller can fail
/// closed rather than seal to a partial recipient set (#47): sealing to a subset
/// while recording the full set as covered permanently locks out the dropped
/// readers. Resolution is local-only, so `did:web`/`did:gitlawb` recipients (and
/// any malformed `did:key`) land in the unresolved set until off-`did:key`
/// resolution exists.
fn resolve_all_recipients(dids: &BTreeSet<String>) -> Result<Vec<VerifyingKey>, Vec<String>> {
let mut keys = Vec::with_capacity(dids.len());
let mut unresolved = Vec::new();
for did in dids {
match did_to_key(did) {
Some(k) => keys.push(k),
None => unresolved.push(did.clone()),
}
}
if unresolved.is_empty() {
Ok(keys)
} else {
Err(unresolved)
}
}
/// What to do with a single withheld blob, decided without any DB or IO so the
/// fail-closed invariant (#47) is unit-testable in isolation.
#[derive(Debug)]
enum SealPlan {
/// An existing envelope already covers exactly this recipient set; nothing to do.
SkipUnchanged,
/// No recipient DID resolved to a key, so there is nothing to seal to.
SkipNoRecipients,
/// At least one recipient DID is unresolvable. Fail closed: never seal to a
/// partial set. Carries the unresolvable DIDs for logging.
SkipUnresolvable(Vec<String>),
/// Seal to `keys` and record coverage under `tag`.
Seal {
keys: Vec<VerifyingKey>,
tag: String,
},
}
/// Decide what to do with one blob given its desired recipient set and the tag
/// already stored for it (if any). Pure: no DB, no IO.
///
/// This is the #47 fail-closed gate in isolation: it returns `Seal` only when
/// EVERY recipient DID resolves, so no caller can seal to a partial set. A
/// changed recipient set (different tag) re-seals so a newly added reader can
/// recover the blob; reader removal is not retroactive (the old envelope is
/// already public). The comparison is on the opaque node-keyed tag, never the
/// DID list.
fn plan_seal(node_seed: &[u8; 32], dids: &BTreeSet<String>, stored_tag: Option<&str>) -> SealPlan {
let tag = recipients_tag(node_seed, dids);
if stored_tag == Some(tag.as_str()) {
return SealPlan::SkipUnchanged;
}
match resolve_all_recipients(dids) {
Ok(keys) if keys.is_empty() => SealPlan::SkipNoRecipients,
Ok(keys) => SealPlan::Seal { keys, tag },
Err(unresolved) => SealPlan::SkipUnresolvable(unresolved),
}
}
/// Encrypt and pin every withheld blob. `recipients` maps blob oid -> DID set;
/// `node_seed` keys the opaque recipients tag. Returns `(oid, cid)` for each blob
/// actually sealed and recorded this call (the per-push delta), used by Option B3
@@ -51,32 +115,53 @@ pub async fn encrypt_and_pin(
recipients: &HashMap<String, BTreeSet<String>>,
) -> Vec<(String, String)> {
let mut sealed = Vec::new();
let mut skipped_unresolvable = 0usize;
for (oid, dids) in recipients {
// Skip only if an existing envelope already covers exactly these
// recipients. If the recipient set changed (e.g. a reader was added to
// the rule), re-seal so the new reader can recover the blob. Reader
// removal is not retroactive: the old envelope is already public. The
// comparison is on the opaque node-keyed tag, never the DID list.
let tag = recipients_tag(node_seed, dids);
match db.encrypted_blob_recipients_tag(repo_id, oid).await {
Ok(Some(stored_tag)) if stored_tag == tag => continue,
Ok(_) => {}
// A DB read failure is not a cache miss: re-sealing here would do an
// avoidable IPFS write during a partial outage. Skip and retry next push.
let stored_tag = match db.encrypted_blob_recipients_tag(repo_id, oid).await {
Ok(t) => t,
Err(e) => {
// A DB read failure is not a cache miss: re-sealing here would do
// an avoidable IPFS write during a partial outage. Skip and retry
// on the next push.
tracing::warn!(oid = %oid, err = %e, "recipients_tag lookup failed; skipping reseal");
continue;
}
}
let keys: Vec<VerifyingKey> = dids.iter().filter_map(|d| did_to_key(d)).collect();
if keys.is_empty() {
tracing::warn!(oid = %oid, "no resolvable recipient keys; skipping encrypted pin");
continue;
}
};
// Fail closed: plan_seal returns Seal only when every recipient DID
// resolves, so we never seal to a partial set and record the full set as
// covered (which would permanently lock out the dropped readers, #47).
let (keys, tag) = match plan_seal(node_seed, dids, stored_tag.as_deref()) {
SealPlan::SkipUnchanged => continue,
SealPlan::SkipNoRecipients => {
tracing::warn!(oid = %oid, "no resolvable recipient keys; skipping encrypted pin");
continue;
}
SealPlan::SkipUnresolvable(unresolved) => {
skipped_unresolvable += 1;
// DIDs are user-controlled (rule reader_dids); log a bounded
// sample, not an unbounded dump. Wording stays neutral about the
// cause: a malformed did:key is not DHT-pending and never will
// resolve, unlike a did:gitlawb awaiting anchoring.
let sample: Vec<&String> = unresolved.iter().take(3).collect();
tracing::warn!(
oid = %oid,
unresolved_count = unresolved.len(),
unresolved_sample = ?sample,
"unresolvable recipient DID(s); skipping encrypted pin to avoid sealing to a partial set"
);
continue;
}
SealPlan::Seal { keys, tag } => (keys, tag),
};
let data = match crate::git::store::read_object(repo_path, oid) {
Ok(Some((_t, bytes))) => bytes,
_ => continue,
Ok(None) => {
tracing::warn!(oid = %oid, "git object not found; skipping encrypted pin");
continue;
}
Err(e) => {
tracing::warn!(oid = %oid, err = %e, "read_object failed; skipping encrypted pin");
continue;
}
};
let envelope = match seal_blob(&data, &keys) {
Ok(e) => e,
@@ -87,7 +172,14 @@ pub async fn encrypt_and_pin(
};
let cid = match crate::ipfs_pin::pin_git_object(ipfs_api, oid, &envelope).await {
Ok(c) if !c.is_empty() => c,
_ => continue,
Ok(_) => {
tracing::warn!(oid = %oid, "pin_git_object returned no cid; skipping");
continue;
}
Err(e) => {
tracing::warn!(oid = %oid, err = %e, "pin_git_object failed; skipping");
continue;
}
};
if let Err(e) = db.record_encrypted_blob(repo_id, oid, &cid, &tag).await {
tracing::warn!(oid = %oid, err = %e, "record_encrypted_blob failed");
@@ -95,16 +187,85 @@ pub async fn encrypt_and_pin(
}
sealed.push((oid.clone(), cid.clone()));
}
// One aggregate signal so a coverage collapse is a single greppable line, not
// a per-oid scrape. In a fully-migrated did:gitlawb org every blob skips and
// recovery coverage silently drops to zero; this is the operator's cue that
// the gap is the deliberate fail-closed posture, not a malfunction.
if skipped_unresolvable > 0 {
tracing::warn!(
sealed = sealed.len(),
skipped = skipped_unresolvable,
"encrypted-pin coverage reduced: blobs skipped for unresolvable recipients"
);
}
sealed
}
#[cfg(test)]
mod tests {
use super::recipients_tag;
use std::collections::BTreeSet;
use super::*;
use ed25519_dalek::SigningKey;
fn set(dids: &[&str]) -> BTreeSet<String> {
dids.iter().map(|s| s.to_string()).collect()
fn did_key(seed: u8) -> String {
let vk = SigningKey::from_bytes(&[seed; 32]).verifying_key();
Did::from_verifying_key(&vk).to_string()
}
// Accepts both `&[String]` (resolve tests, built from `did_key`) and
// `&[&str]` (tag tests, built from literals).
fn set<S: AsRef<str>>(dids: &[S]) -> BTreeSet<String> {
dids.iter().map(|s| s.as_ref().to_string()).collect()
}
#[test]
fn all_did_key_recipients_resolve() {
let dids = set(&[did_key(1), did_key(2)]);
let keys = resolve_all_recipients(&dids).expect("all did:key resolve");
assert_eq!(keys.len(), 2);
}
#[test]
fn mixed_set_with_did_gitlawb_fails_closed() {
// Core #47 regression: a resolvable subset must never yield a sealable
// key set. Two did:key plus one did:gitlawb -> Err naming only the
// unresolvable DID.
let gitlawb = Did::gitlawb("zSomeDhtKey").to_string();
let dids = set(&[did_key(1), did_key(2), gitlawb.clone()]);
let unresolved = resolve_all_recipients(&dids).expect_err("must fail closed");
assert_eq!(unresolved, vec![gitlawb]);
}
#[test]
fn did_web_recipient_fails_closed() {
let web = Did::web("example.com").to_string();
let dids = set(&[did_key(1), web.clone()]);
let unresolved = resolve_all_recipients(&dids).expect_err("did:web cannot resolve locally");
assert_eq!(unresolved, vec![web]);
}
#[test]
fn malformed_did_key_fails_closed() {
// The third state: not a method-resolution gap but a permanently broken
// did:key (invalid multibase). Must also fail closed.
let broken = "did:key:z!!!invalid".to_string();
let dids = set(&[did_key(1), broken.clone()]);
let unresolved = resolve_all_recipients(&dids).expect_err("malformed did:key fails");
assert_eq!(unresolved, vec![broken]);
}
#[test]
fn empty_set_resolves_to_empty_keys() {
let dids = BTreeSet::new();
let keys = resolve_all_recipients(&dids).expect("empty set is not an error");
assert!(keys.is_empty());
}
#[test]
fn single_unresolvable_did_returns_that_did() {
let gitlawb = Did::gitlawb("zOnlyOne").to_string();
let dids: BTreeSet<String> = [gitlawb.clone()].into_iter().collect();
let unresolved = resolve_all_recipients(&dids).expect_err("must fail closed");
assert_eq!(unresolved, vec![gitlawb]);
}
#[test]
@@ -133,4 +294,69 @@ mod tests {
"tag must depend on the node seed, not be a plain hash"
);
}
// plan_seal is the seal/skip decision `encrypt_and_pin` acts on. Testing it
// directly pins the #47 fail-closed invariant at the function that owns it,
// which a unit test of `resolve_all_recipients` alone cannot do (it can't
// catch the caller falling through to a partial seal).
const SEED: [u8; 32] = [9u8; 32];
#[test]
fn plan_seal_seals_when_all_recipients_resolve() {
let dids = set(&[did_key(1), did_key(2)]);
match plan_seal(&SEED, &dids, None) {
SealPlan::Seal { keys, tag } => {
assert_eq!(keys.len(), 2, "must seal to the full recipient set");
assert_eq!(
tag,
recipients_tag(&SEED, &dids),
"records the full-set tag"
);
}
other => panic!("expected Seal, got {other:?}"),
}
}
#[test]
fn plan_seal_fails_closed_on_any_unresolvable_recipient() {
// The #47 invariant at the decision boundary: one unresolvable DID among
// resolvable ones must NOT yield a Seal (which would seal a partial set).
let gitlawb = Did::gitlawb("zPending").to_string();
let dids = set(&[did_key(1), did_key(2), gitlawb.clone()]);
match plan_seal(&SEED, &dids, None) {
SealPlan::SkipUnresolvable(unresolved) => assert_eq!(unresolved, vec![gitlawb]),
other => panic!("must fail closed, never seal a partial set; got {other:?}"),
}
}
#[test]
fn plan_seal_skips_empty_recipient_set() {
let dids = BTreeSet::new();
assert!(matches!(
plan_seal(&SEED, &dids, None),
SealPlan::SkipNoRecipients
));
}
#[test]
fn plan_seal_skips_when_tag_unchanged() {
let dids = set(&[did_key(1)]);
let stored = recipients_tag(&SEED, &dids);
assert!(matches!(
plan_seal(&SEED, &dids, Some(&stored)),
SealPlan::SkipUnchanged
));
}
#[test]
fn plan_seal_reseals_when_recipient_set_changed() {
// A stored tag for a DIFFERENT set is a miss: a newly added reader must
// trigger a re-seal, not be skipped as unchanged.
let dids = set(&[did_key(1), did_key(2)]);
let stale = recipients_tag(&SEED, &set(&[did_key(1)]));
match plan_seal(&SEED, &dids, Some(&stale)) {
SealPlan::Seal { keys, .. } => assert_eq!(keys.len(), 2),
other => panic!("changed recipient set must re-seal; got {other:?}"),
}
}
}
+12
View File
@@ -23,6 +23,9 @@ pub enum AppError {
#[allow(dead_code)]
Forbidden(String),
#[error("icaptcha proof required: {0}")]
IcaptchaProofRequired(String),
#[error("invalid request: {0}")]
BadRequest(String),
@@ -52,6 +55,15 @@ impl IntoResponse for AppError {
AppError::NotFound(msg) => (StatusCode::NOT_FOUND, "not_found", msg.clone()),
AppError::Unauthorized(msg) => (StatusCode::UNAUTHORIZED, "not_an_agent", msg.clone()),
AppError::Forbidden(msg) => (StatusCode::FORBIDDEN, "forbidden", msg.clone()),
// 403, not 401: the caller IS an authenticated agent (credentials are
// valid) but is forbidden from this action without a valid, fresh
// iCaptcha proof. The distinct `icaptcha_proof_required` code — which
// clients branch on — keeps it separable from a plain `forbidden`.
AppError::IcaptchaProofRequired(msg) => (
StatusCode::FORBIDDEN,
"icaptcha_proof_required",
msg.clone(),
),
AppError::BadRequest(msg) => (StatusCode::BAD_REQUEST, "bad_request", msg.clone()),
AppError::Git(msg) => (StatusCode::INTERNAL_SERVER_ERROR, "git_error", msg.clone()),
AppError::Db(e) => (StatusCode::INTERNAL_SERVER_ERROR, "db_error", e.to_string()),
+11 -6
View File
@@ -23,13 +23,18 @@ pub fn create_issue(repo_path: &Path, issue_id: &str, json: &str) -> Result<()>
use std::io::Write;
let mut child = hash_output;
if let Some(stdin) = child.stdin.take() {
let mut stdin = stdin;
stdin
.write_all(json.as_bytes())
.context("failed to write to git hash-object stdin")?;
}
// Write the blob to stdin, but always reap the child afterward even if the
// write fails, so a stdin-write error can't drop the Child unwaited and leak
// a zombie (#53).
let write_result = match child.stdin.take() {
Some(mut stdin) => stdin.write_all(json.as_bytes()),
None => Err(std::io::Error::new(
std::io::ErrorKind::BrokenPipe,
"git hash-object stdin unavailable",
)),
};
let output = child.wait_with_output().context("git hash-object failed")?;
write_result.context("failed to write to git hash-object stdin")?;
if !output.status.success() {
let stderr = String::from_utf8_lossy(&output.stderr);
+1
View File
@@ -1,4 +1,5 @@
pub mod issues;
pub mod push_delta;
pub mod repo_store;
pub mod smart_http;
pub mod store;
+663
View File
@@ -0,0 +1,663 @@
//! Push-delta object enumeration for the IPFS/Pinata pin path.
//!
//! The pin path needs the set of git objects to consider for pinning after a
//! push. Historically both pin functions enumerated the *whole* repo
//! (`git cat-file --batch-all-objects`) on every push, so cost scaled with
//! repo size rather than push size (finding N16). This module computes the
//! per-push *delta* instead — the objects a push introduced — and keeps the
//! whole-repo lister for the reconciliation sweep and the fail-closed fallback.
//!
//! ## Correctness framing (do not confuse with the #84 withholding direction)
//!
//! The pin enumeration is part of the *exposure* set, not the withheld filter.
//! On the *delta* path, narrowing per push only *shrinks* what is pinned, so it
//! cannot create an under-withholding leak — the reachable-only withheld filter
//! (`visibility_pack`) still subtracts from the smaller set we feed it. The only
//! risk there is *under-pinning* (a durability gap) the sweep backstops.
//!
//! The *full-scan* path is different: `list_all_objects` includes dangling
//! objects the reachable withheld set never classified, so subtracting that set
//! is not enough — a dangling private blob would slip through (#99). The caller
//! signals a full scan via [`PinCandidateSet::full_scan`] and must then apply
//! the fail-closed blob filter (`visibility_pack::replicable_objects_fail_closed`)
//! instead of the plain reachable-only subtraction.
//!
//! Because the pin candidate set needs only the OID *set* (never the per-path
//! information the withheld classifier needs), `git rev-list --objects
//! --no-object-names` is safe here. The "rev-list reports one path per object"
//! trap from #84 applies only to `visibility_pack`'s per-path `ls-tree` walk.
use std::collections::HashSet;
use std::path::{Path, PathBuf};
use std::process::Command;
use anyhow::{bail, Context, Result};
/// Env var that forces the push path to always full-scan, bypassing the delta
/// optimization (KTD7 kill-switch). Reuses the already-tested fallback branch
/// rather than introducing a second enumeration path. Does not gate the sweep.
const FORCE_FULL_SCAN_ENV: &str = "GITLAWB_PIN_FORCE_FULLSCAN";
/// The objects a push introduced, or a signal that the delta could not be
/// computed safely and the caller should fall back to a full scan.
///
/// Three-valued by design: `Delta(vec![])` (a no-op push — pin nothing) is a
/// *different* fact from `FullScanRequired` (could not compute — fall back),
/// and they drive different actions.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum PinCandidates {
/// Objects reachable from the new tips but not the old tips.
Delta(Vec<String>),
/// The delta is not safe to use; the caller must enumerate the whole repo
/// with [`list_all_objects`] instead. Returned on a non-commit tip, any git
/// error, or the force-full-scan kill-switch.
FullScanRequired,
}
/// Resolve the set of objects introduced by a push from its ref updates.
///
/// `new_tips` / `old_tips` are the non-zero new/old SHAs of the push's ref
/// updates (the caller strips the all-zeros create/delete sentinel). Keeping
/// the helper on plain SHA slices (not the `api` `RefUpdate` type) keeps it
/// unit-testable without the HTTP layer.
///
/// Fail-closed: any condition where the introduced set cannot be safely
/// determined returns [`PinCandidates::FullScanRequired`] rather than a partial
/// set, so the caller full-scans instead of silently under-pinning.
pub fn resolve_push_delta(repo_path: &Path, new_tips: &[&str], old_tips: &[&str]) -> PinCandidates {
// KTD7 kill-switch: force the (already-tested) full-scan fallback. The env
// read is split out from the pure logic so the resolver stays unit-testable
// without touching process-global state.
resolve_push_delta_inner(repo_path, new_tips, old_tips, force_full_scan())
}
/// Whether the force-full-scan kill-switch env var is set.
fn force_full_scan() -> bool {
std::env::var_os(FORCE_FULL_SCAN_ENV).is_some()
}
// Intentionally private but reachable from the test module (via `use super::*`)
// so the kill-switch can be exercised without mutating process-global env.
fn resolve_push_delta_inner(
repo_path: &Path,
new_tips: &[&str],
old_tips: &[&str],
force_full_scan: bool,
) -> PinCandidates {
if force_full_scan {
tracing::debug!("{FORCE_FULL_SCAN_ENV} set — forcing full scan");
return PinCandidates::FullScanRequired;
}
// A push that only deleted refs introduces no objects.
if new_tips.is_empty() {
return PinCandidates::Delta(Vec::new());
}
// Ref-type guard (the fail-closed mechanism). `git rev-list` does NOT error
// on a non-commit tip — a blob/tree/tag-of-non-commit all exit 0 and walk
// the object — so the rev-list exit code cannot catch this. Check each new
// tip's *fully-peeled* type with `cat-file -t '<sha>^{}'` (recursive peel:
// tag-of-commit -> commit, tag-of-tree -> tree, tag-of-tag-of-commit ->
// commit). Bare `cat-file -t` returns `tag` for an annotated tag, and
// `for-each-ref %(*objecttype)` peels only one level — neither is correct.
for tip in new_tips {
match peeled_object_type(repo_path, tip) {
Some(t) if t == "commit" => {}
other => {
tracing::debug!(
tip = %tip,
peeled_type = ?other,
"push tip is not a commit (or type lookup failed) — forcing full scan"
);
return PinCandidates::FullScanRequired;
}
}
}
match rev_list_delta(repo_path, new_tips, old_tips) {
Ok(oids) => PinCandidates::Delta(oids),
Err(e) => {
tracing::debug!(err = %e, "push-delta rev-list failed — forcing full scan");
PinCandidates::FullScanRequired
}
}
}
/// Return the fully-peeled object type of `sha` (e.g. `commit`, `tree`,
/// `blob`), or `None` if the object is missing/unpeelable or git errored.
fn peeled_object_type(repo_path: &Path, sha: &str) -> Option<String> {
let output = Command::new("git")
.args(["cat-file", "-t", &format!("{sha}^{{}}")])
.current_dir(repo_path)
.output()
.ok()?;
if !output.status.success() {
return None;
}
Some(String::from_utf8_lossy(&output.stdout).trim().to_string())
}
/// Run `git rev-list --objects --no-object-names <new> --not <old>` and return
/// the bare OID set. Decides on `status.success()` *before* parsing stdout, so
/// a walk that prints a valid prefix then errors mid-walk is discarded.
fn rev_list_delta(repo_path: &Path, new_tips: &[&str], old_tips: &[&str]) -> Result<Vec<String>> {
let mut args: Vec<&str> = vec!["rev-list", "--objects", "--no-object-names"];
args.extend_from_slice(new_tips);
if !old_tips.is_empty() {
args.push("--not");
args.extend_from_slice(old_tips);
}
let output = Command::new("git")
.args(&args)
.current_dir(repo_path)
.output()
.context("failed to run git rev-list for push delta")?;
if !output.status.success() {
let stderr = String::from_utf8_lossy(&output.stderr);
bail!("git rev-list failed: {stderr}");
}
let stdout = String::from_utf8_lossy(&output.stdout);
Ok(stdout
.lines()
.map(|l| l.trim().to_string())
.filter(|l| !l.is_empty())
.collect())
}
/// List every object in the repository via
/// `git cat-file --batch-all-objects --batch-check='%(objectname)'`.
///
/// This is the whole-repo enumeration the push path falls back to and the
/// reconciliation sweep relies on. It returns *all* objects (including
/// unreachable/dangling ones), which is what the sweep needs to catch
/// stragglers — do not swap it for a reachability walk.
pub fn list_all_objects(repo_path: &Path) -> Result<Vec<String>> {
let output = Command::new("git")
.args([
"cat-file",
"--batch-all-objects",
"--batch-check=%(objectname)",
])
.current_dir(repo_path)
.output()
.context("failed to run git cat-file")?;
if !output.status.success() {
let stderr = String::from_utf8_lossy(&output.stderr);
bail!("git cat-file failed: {stderr}");
}
let stdout = String::from_utf8_lossy(&output.stdout);
Ok(stdout
.lines()
.map(|l| l.trim().to_string())
.filter(|l| !l.is_empty())
.collect())
}
/// Like [`list_all_objects`] but pairs each OID with its object type, via
/// `--batch-check='%(objectname) %(objecttype)'`. The pin path's fail-closed
/// filter needs to tell blobs (content, withholdable) from commits/trees
/// (structural, never withheld) without typing the candidate list itself.
pub fn list_all_objects_with_type(repo_path: &Path) -> Result<Vec<(String, String)>> {
let output = Command::new("git")
.args([
"cat-file",
"--batch-all-objects",
"--batch-check=%(objectname) %(objecttype)",
])
.current_dir(repo_path)
.output()
.context("failed to run git cat-file")?;
if !output.status.success() {
let stderr = String::from_utf8_lossy(&output.stderr);
bail!("git cat-file failed: {stderr}");
}
let stdout = String::from_utf8_lossy(&output.stdout);
Ok(stdout
.lines()
.filter_map(|l| {
let mut it = l.split_whitespace();
let oid = it.next()?;
let ty = it.next()?;
Some((oid.to_string(), ty.to_string()))
})
.collect())
}
/// Every blob OID in the repo, including unreachable/dangling ones. The
/// fail-closed pin filter drops any candidate blob absent from the reachable,
/// visibility-allowed set; a dangling private blob is in this set but not the
/// allowed set, so it never replicates (#99).
pub fn all_blob_oids(repo_path: &Path) -> Result<HashSet<String>> {
Ok(list_all_objects_with_type(repo_path)?
.into_iter()
.filter(|(_, ty)| ty == "blob")
.map(|(oid, _)| oid)
.collect())
}
/// The pin candidate OIDs for a push plus whether they came from a full-repo
/// scan. `full_scan` is true when the delta could not be used and the whole
/// object DB (including dangling objects) was enumerated — the caller must then
/// apply the fail-closed blob filter, because the reachable-only withheld set
/// cannot classify dangling blobs (#99).
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct PinCandidateSet {
pub candidates: Vec<String>,
pub full_scan: bool,
}
/// Resolve the pin candidate OID set for a push, off the async worker.
///
/// Runs [`resolve_push_delta`] in `spawn_blocking` (git subprocess) and applies
/// the full-scan fallback for [`PinCandidates::FullScanRequired`]. Owns the
/// dispatch so the push call site stays thin and this wiring is unit-testable
/// without the HTTP layer.
///
/// Every degraded path is **logged**, not silent: a full-scan fallback, a
/// failed full scan, and a panicked blocking task each emit a warning. On a
/// failed full scan or a task panic the candidate set is empty (pin nothing
/// this push); that is a durability gap the reconciliation sweep backstops, and
/// it can never leak because the withheld/fail-closed filter still runs on
/// whatever set is returned. `full_scan` rides on the returned set so the caller
/// knows when the dangling-inclusive filter is required.
pub async fn resolve_candidates_for_push(
repo_path: PathBuf,
new_tips: Vec<String>,
old_tips: Vec<String>,
) -> PinCandidateSet {
tokio::task::spawn_blocking(move || {
let new_refs: Vec<&str> = new_tips.iter().map(String::as_str).collect();
let old_refs: Vec<&str> = old_tips.iter().map(String::as_str).collect();
match resolve_push_delta(&repo_path, &new_refs, &old_refs) {
PinCandidates::Delta(objs) => {
tracing::info!(delta = objs.len(), repo = %repo_path.display(), "pin candidate set from push delta");
PinCandidateSet { candidates: objs, full_scan: false }
}
PinCandidates::FullScanRequired => {
tracing::warn!(repo = %repo_path.display(), "pin delta unavailable (non-commit tip, git error, or force-full-scan) — full-scan fallback");
match list_all_objects(&repo_path) {
Ok(objs) => PinCandidateSet { candidates: objs, full_scan: true },
Err(e) => {
tracing::warn!(repo = %repo_path.display(), err = %e, "full-scan fallback failed; pinning nothing this push (reconciliation sweep backstops)");
PinCandidateSet { candidates: Vec::new(), full_scan: false }
}
}
}
}
})
.await
.unwrap_or_else(|e| {
tracing::warn!(err = %e, "pin candidate computation task panicked; pinning nothing this push (reconciliation sweep backstops)");
PinCandidateSet { candidates: Vec::new(), full_scan: false }
})
}
#[cfg(test)]
mod tests {
use super::*;
use std::collections::HashSet;
use tempfile::TempDir;
/// Minimal git helper for building test repos.
struct Repo {
_td: TempDir,
path: std::path::PathBuf,
}
impl Repo {
fn new() -> Self {
let td = TempDir::new().unwrap();
let path = td.path().to_path_buf();
let r = Repo { _td: td, path };
r.git(&["init", "-q", "-b", "main"]);
r.git(&["config", "user.email", "t@t"]);
r.git(&["config", "user.name", "t"]);
r
}
fn git(&self, args: &[&str]) -> String {
let out = Command::new("git")
.args(args)
.current_dir(&self.path)
.output()
.unwrap();
assert!(
out.status.success(),
"git {args:?} failed: {}",
String::from_utf8_lossy(&out.stderr)
);
String::from_utf8_lossy(&out.stdout).trim().to_string()
}
fn commit_file(&self, name: &str, body: &str) -> String {
std::fs::write(self.path.join(name), body).unwrap();
self.git(&["add", name]);
self.git(&["commit", "-qm", &format!("add {name}")]);
self.git(&["rev-parse", "HEAD"])
}
fn rev(&self, r: &str) -> String {
self.git(&["rev-parse", r])
}
}
const ZERO: &str = "0000000000000000000000000000000000000000";
fn delta(c: PinCandidates) -> Vec<String> {
match c {
PinCandidates::Delta(v) => v,
PinCandidates::FullScanRequired => panic!("expected Delta, got FullScanRequired"),
}
}
#[test]
fn single_new_commit_delta_excludes_preexisting_objects() {
let repo = Repo::new();
let c1 = repo.commit_file("a.txt", "one\n");
let c2 = repo.commit_file("b.txt", "two\n");
let got: HashSet<String> = delta(resolve_push_delta(&repo.path, &[&c2], &[&c1]))
.into_iter()
.collect();
// The new blob b.txt and commit c2 are in the delta; the old blob a.txt
// and commit c1 are not.
let new_blob = repo.rev("HEAD:b.txt");
let old_blob = repo.rev(&format!("{c1}:a.txt"));
assert!(got.contains(&c2), "new commit in delta");
assert!(got.contains(&new_blob), "new blob in delta");
assert!(!got.contains(&c1), "old commit excluded");
assert!(!got.contains(&old_blob), "old blob excluded");
}
#[test]
fn created_ref_is_superset_of_new_objects() {
// A created ref (old tip all-zeros, filtered out by caller) with no old
// tips walks everything reachable from the new tip — a superset of the
// genuinely new objects, never fewer.
let repo = Repo::new();
let c1 = repo.commit_file("a.txt", "one\n");
let got: HashSet<String> = delta(resolve_push_delta(&repo.path, &[&c1], &[]))
.into_iter()
.collect();
assert!(got.contains(&c1));
assert!(got.contains(&repo.rev("HEAD:a.txt")));
assert!(got.contains(&repo.rev("HEAD^{tree}")));
}
#[test]
fn force_push_delta_is_objects_unique_to_new_tip() {
let repo = Repo::new();
let base = repo.commit_file("a.txt", "one\n");
let old_tip = repo.commit_file("b.txt", "two\n");
// Rewrite history: reset to base, commit a different file.
repo.git(&["reset", "-q", "--hard", &base]);
let new_tip = repo.commit_file("c.txt", "three\n");
let got: HashSet<String> = delta(resolve_push_delta(&repo.path, &[&new_tip], &[&old_tip]))
.into_iter()
.collect();
assert!(got.contains(&new_tip), "new tip in delta");
assert!(got.contains(&repo.rev(&format!("{new_tip}:c.txt"))));
// No error; force-push computes new-minus-old cleanly.
}
#[test]
fn deleted_ref_only_yields_empty_delta_without_git() {
let repo = Repo::new();
repo.commit_file("a.txt", "one\n");
// All updates were deletions => new_tips empty after the caller strips zeros.
assert_eq!(
resolve_push_delta(&repo.path, &[], &[ZERO]),
PinCandidates::Delta(Vec::new())
);
}
#[test]
fn blob_tip_forces_full_scan() {
let repo = Repo::new();
repo.commit_file("a.txt", "one\n");
let blob = repo.rev("HEAD:a.txt");
assert_eq!(
resolve_push_delta(&repo.path, &[&blob], &[]),
PinCandidates::FullScanRequired,
"a blob tip must force full scan (rev-list would exit 0 and walk it)"
);
}
#[test]
fn tree_tip_forces_full_scan() {
let repo = Repo::new();
repo.commit_file("a.txt", "one\n");
let tree = repo.rev("HEAD^{tree}");
assert_eq!(
resolve_push_delta(&repo.path, &[&tree], &[]),
PinCandidates::FullScanRequired
);
}
#[test]
fn annotated_tag_of_tree_forces_full_scan() {
let repo = Repo::new();
repo.commit_file("a.txt", "one\n");
let tree = repo.rev("HEAD^{tree}");
repo.git(&["tag", "-a", "treetag", "-m", "x", &tree]);
let tag = repo.rev("treetag");
assert_eq!(
resolve_push_delta(&repo.path, &[&tag], &[]),
PinCandidates::FullScanRequired,
"annotated tag peeling to a tree must force full scan"
);
}
#[test]
fn tag_of_tag_of_noncommit_forces_full_scan() {
let repo = Repo::new();
repo.commit_file("a.txt", "one\n");
let blob = repo.rev("HEAD:a.txt");
repo.git(&["tag", "-a", "t1", "-m", "x", &blob]);
let t1 = repo.rev("t1");
repo.git(&["tag", "-a", "t2", "-m", "x", &t1]);
let t2 = repo.rev("t2");
assert_eq!(
resolve_push_delta(&repo.path, &[&t2], &[]),
PinCandidates::FullScanRequired
);
}
#[test]
fn annotated_tag_of_commit_proceeds_as_delta() {
// DISCRIMINATING positive-path regression: an annotated tag whose peeled
// target is a commit must return Delta, not FullScanRequired. A
// FullScanRequired here means the guard used bare `cat-file -t` (returns
// `tag`) instead of the `^{}` peel.
let repo = Repo::new();
let c1 = repo.commit_file("a.txt", "one\n");
repo.git(&["tag", "-a", "rel", "-m", "release", &c1]);
let tag = repo.rev("rel");
let got: HashSet<String> = delta(resolve_push_delta(&repo.path, &[&tag], &[]))
.into_iter()
.collect();
assert!(
got.contains(&c1),
"peeled commit's objects are in the delta"
);
}
#[test]
fn tag_of_tag_of_commit_proceeds_as_delta() {
// Deep-peel positive path: tag -> tag -> commit must recurse to `commit`
// via `^{}` and return Delta. Complements tag_of_tag_of_noncommit (which
// forces full scan) — guards against a regression in peel depth that
// would silently downgrade a legitimate nested-tag push to full scan.
let repo = Repo::new();
let c1 = repo.commit_file("a.txt", "one\n");
repo.git(&["tag", "-a", "t1", "-m", "x", &c1]);
let t1 = repo.rev("t1");
repo.git(&["tag", "-a", "t2", "-m", "x", &t1]);
let t2 = repo.rev("t2");
let got: HashSet<String> = delta(resolve_push_delta(&repo.path, &[&t2], &[]))
.into_iter()
.collect();
assert!(
got.contains(&c1),
"peeled commit's objects are in the delta"
);
}
#[tokio::test]
async fn resolve_candidates_for_push_returns_delta() {
// The extracted wiring returns the per-push delta on the happy path.
let repo = Repo::new();
let c1 = repo.commit_file("a.txt", "one\n");
let c2 = repo.commit_file("b.txt", "two\n");
let set =
resolve_candidates_for_push(repo.path.clone(), vec![c2.clone()], vec![c1.clone()])
.await;
assert!(!set.full_scan, "happy-path delta is not a full scan");
let got: HashSet<String> = set.candidates.into_iter().collect();
let new_blob = repo.rev("HEAD:b.txt");
assert!(
got.contains(&c2) && got.contains(&new_blob),
"new objects pinned"
);
assert!(!got.contains(&c1), "old commit excluded from delta");
}
#[tokio::test]
async fn resolve_candidates_for_push_falls_back_to_full_scan_on_noncommit_tip() {
// A non-commit tip forces FullScanRequired, and the wiring resolves that
// to the whole-repo list (a superset), never an empty set.
let repo = Repo::new();
repo.commit_file("a.txt", "one\n");
let blob = repo.rev("HEAD:a.txt");
let all: HashSet<String> = list_all_objects(&repo.path).unwrap().into_iter().collect();
let set = resolve_candidates_for_push(repo.path.clone(), vec![blob], vec![]).await;
assert!(set.full_scan, "non-commit tip is signalled as a full scan");
let got: HashSet<String> = set.candidates.into_iter().collect();
assert_eq!(got, all, "non-commit tip falls back to full repo scan");
}
#[test]
fn missing_oid_tip_forces_full_scan() {
let repo = Repo::new();
repo.commit_file("a.txt", "one\n");
let bogus = "deadbeefdeadbeefdeadbeefdeadbeefdeadbeef";
assert_eq!(
resolve_push_delta(&repo.path, &[bogus], &[]),
PinCandidates::FullScanRequired,
"a missing/corrupt tip OID must force full scan"
);
}
#[test]
fn present_noncommit_old_tip_is_safe() {
// A present-but-non-commit old_sha (a --not arg) must not under-pin:
// either rev-list errors (FullScanRequired) or it over-lists (Delta).
// Either way the new tip's objects are covered; never fewer.
let repo = Repo::new();
let c1 = repo.commit_file("a.txt", "one\n");
let c2 = repo.commit_file("b.txt", "two\n");
let old_tree = repo.rev(&format!("{c1}^{{tree}}"));
let result = resolve_push_delta(&repo.path, &[&c2], &[&old_tree]);
match result {
PinCandidates::FullScanRequired => {} // safe: caller full-scans
PinCandidates::Delta(objs) => {
// Safe direction: must still contain the new commit (over-list ok).
assert!(objs.contains(&c2), "new tip objects must be covered");
}
}
}
#[test]
fn multi_ref_push_unions_ranges() {
let repo = Repo::new();
let base = repo.commit_file("a.txt", "one\n");
// branch1 advances
let b1 = repo.commit_file("b.txt", "two\n");
// branch2 from base advances independently
repo.git(&["checkout", "-q", "-b", "branch2", &base]);
let b2 = repo.commit_file("c.txt", "three\n");
let got: HashSet<String> =
delta(resolve_push_delta(&repo.path, &[&b1, &b2], &[&base, &base]))
.into_iter()
.collect();
assert!(got.contains(&b1), "branch1 new commit");
assert!(got.contains(&b2), "branch2 new commit");
}
#[test]
fn empty_repo_no_tips_yields_empty_delta() {
let repo = Repo::new();
assert_eq!(
resolve_push_delta(&repo.path, &[], &[]),
PinCandidates::Delta(Vec::new())
);
}
#[test]
fn force_full_scan_flag_overrides() {
// Test the kill-switch via the pure inner fn so we never touch
// process-global env (which would race with parallel tests, and is
// unsafe in Rust 2024). The public resolve_push_delta wires the env
// read into this same flag.
let repo = Repo::new();
let c1 = repo.commit_file("a.txt", "one\n");
assert_eq!(
resolve_push_delta_inner(&repo.path, &[&c1], &[], true),
PinCandidates::FullScanRequired
);
// And with the flag off, the same push yields a Delta.
assert!(matches!(
resolve_push_delta_inner(&repo.path, &[&c1], &[], false),
PinCandidates::Delta(_)
));
}
#[test]
fn list_all_objects_returns_full_repo() {
let repo = Repo::new();
repo.commit_file("a.txt", "one\n");
repo.commit_file("b.txt", "two\n");
let all = list_all_objects(&repo.path).unwrap();
// 2 commits + 2 trees + 2 blobs = 6 objects.
assert_eq!(all.len(), 6, "got: {all:?}");
}
#[test]
fn all_blob_oids_includes_dangling_and_excludes_nonblobs() {
// The all-blob universe must include a dangling blob (the #99 hazard) and
// exclude commits/trees, so the fail-closed filter can tell a blob from a
// structural object without typing the candidate list.
let repo = Repo::new();
let c1 = repo.commit_file("a.txt", "one\n");
let reachable_blob = repo.rev("HEAD:a.txt");
let tree = repo.rev("HEAD^{tree}");
// Write a blob referenced by no tree/commit — dangling.
std::fs::write(repo.path.join("orphan.bin"), b"dangling\n").unwrap();
let dangling = repo.git(&["hash-object", "-w", "orphan.bin"]);
let blobs = all_blob_oids(&repo.path).unwrap();
assert!(blobs.contains(&reachable_blob), "reachable blob present");
assert!(
blobs.contains(&dangling),
"dangling blob present in the all-blob universe"
);
assert!(!blobs.contains(&c1), "commit is not a blob");
assert!(!blobs.contains(&tree), "tree is not a blob");
// The typed lister tags each object; spot-check the dangling blob's type.
let typed = list_all_objects_with_type(&repo.path).unwrap();
assert!(
typed
.iter()
.any(|(oid, ty)| oid == &dangling && ty == "blob"),
"dangling object is typed as a blob"
);
}
}
+200 -12
View File
@@ -75,23 +75,83 @@ pub async fn receive_pack(repo_path: &Path, request_body: Bytes) -> Result<Respo
.body(Body::from(output))?)
}
/// Sends SIGTERM to a child's whole process group on drop, unless disarmed first.
///
/// A served `git upload-pack`/`receive-pack` forks helpers such as `pack-objects`.
/// If the request future is dropped (client disconnect) or returns early, dropping
/// the tokio `Child` does not signal `git`; it lingers until EPIPE, and its
/// `pack-objects` child can reparent to PID 1 and never be reaped — a zombie that
/// accumulates until `fork()` fails with EAGAIN (#53). Spawning the child in its
/// own process group and signalling that group here tears the whole tree down at
/// the source. SIGTERM (not SIGKILL) lets `git` run its cleanup — notably removing
/// `.git/*.lock` files mid-`receive-pack` — before it exits, so an aborted push
/// can't leave a stale lock that blocks the next one. The guard is disarmed once
/// `wait_with_output()` returns, so a request that completed cleanly never signals.
#[cfg(unix)]
struct KillGroupOnDrop {
pgid: Option<i32>,
}
#[cfg(unix)]
impl KillGroupOnDrop {
fn disarm(&mut self) {
self.pgid = None;
}
}
#[cfg(unix)]
impl Drop for KillGroupOnDrop {
fn drop(&mut self) {
if let Some(pgid) = self.pgid {
// SAFETY: kill(2) takes only integer arguments and borrows no Rust
// memory. Signalling a stale group just returns ESRCH, which we ignore.
unsafe {
libc::kill(-pgid, libc::SIGTERM);
}
}
}
}
async fn run_git_service(service: &str, repo_path: &Path, input: Bytes) -> Result<Vec<u8>> {
let mut child = Command::new("git")
let mut command = Command::new("git");
command
.arg(service_to_command(service))
.arg("--stateless-rpc")
.arg(repo_path)
.stdin(Stdio::piped())
.stdout(Stdio::piped())
.stderr(Stdio::piped())
.spawn()?;
.stderr(Stdio::piped());
// Write request body to git's stdin
if let Some(mut stdin) = child.stdin.take() {
stdin.write_all(&input).await?;
}
// Run git in its own process group so the whole tree (git + pack-objects)
// can be signalled together on disconnect rather than orphaning a grandchild.
#[cfg(unix)]
command.process_group(0);
let mut child = command.spawn()?;
// Arm the group-kill guard for the lifetime of the request. With
// process_group(0) the child is its own group leader, so pgid == its pid.
#[cfg(unix)]
let mut group_guard = KillGroupOnDrop {
pgid: child.id().map(|id| id as i32),
};
// Write the request body to git's stdin, but don't early-return on a write
// error: always reap the child first (below), so the guard only ever fires on
// an actual future-drop (client disconnect), never on a pid we just reaped.
let write_result: std::io::Result<()> = match child.stdin.take() {
Some(mut stdin) => stdin.write_all(&input).await,
None => Ok(()),
};
let output = child.wait_with_output().await?;
// Child reaped, so its group is gone: disarm before surfacing any error.
#[cfg(unix)]
group_guard.disarm();
write_result.context("failed to write to git stdin")?;
if !output.status.success() {
let stderr = String::from_utf8_lossy(&output.stderr);
bail!("{service} failed: {stderr}");
@@ -152,13 +212,25 @@ pub fn build_filtered_pack(repo_path: &Path, withheld: &HashSet<String>) -> Resu
.stdout(Stdio::piped())
.stderr(Stdio::piped())
.spawn()?;
{
// Feed the object ids on stdin, but always reap the child afterward even if
// the write fails or stdin is missing, so an error can't drop the Child
// unwaited and leak a zombie (#53).
let write_result: std::io::Result<()> = {
use std::io::Write as _;
let mut stdin = child.stdin.take().expect("stdin");
stdin.write_all(keep.join("\n").as_bytes())?;
stdin.write_all(b"\n")?;
}
match child.stdin.take() {
Some(mut stdin) => {
let mut data = keep.join("\n").into_bytes();
data.push(b'\n');
stdin.write_all(&data)
}
None => Err(std::io::Error::new(
std::io::ErrorKind::BrokenPipe,
"git pack-objects stdin unavailable",
)),
}
};
let out = child.wait_with_output()?;
write_result.context("failed to write object ids to git pack-objects stdin")?;
if !out.status.success() {
bail!(
"git pack-objects failed: {}",
@@ -688,4 +760,120 @@ mod tests {
server.abort();
}
// ── #53 regression: served-git process-group teardown ──────────────────
//
// run_git_service runs git in its own process group and SIGTERMs that group
// when the request future is dropped (client disconnect) or returns early, so
// git AND its pack-objects child die together instead of orphaning a zombie.
// These exercise the KillGroupOnDrop guard that wires that up.
#[cfg(unix)]
fn alive(pid: i32) -> bool {
// kill(pid, 0) probes existence: returns 0 while the pid exists, -1 once
// it's gone (ESRCH). Same-uid here, so EPERM never applies.
unsafe { libc::kill(pid, 0) == 0 }
}
#[cfg(unix)]
#[tokio::test]
async fn kill_group_guard_terminates_child_on_drop() {
let mut child = tokio::process::Command::new("sleep")
.arg("300")
.process_group(0)
.spawn()
.unwrap();
let pgid = child.id().map(|id| id as i32);
{
let _guard = KillGroupOnDrop { pgid };
} // guard drops here -> SIGTERM to the group
use std::os::unix::process::ExitStatusExt;
let status = child.wait().await.unwrap();
assert_eq!(
status.signal(),
Some(libc::SIGTERM),
"child must be terminated by SIGTERM via its process group"
);
}
#[cfg(unix)]
#[tokio::test]
async fn kill_group_guard_reaps_grandchild_on_drop() {
// The #53 scenario: git forks pack-objects. A group kill must reach the
// grandchild, not just the direct child. sh (the group leader) backgrounds
// a sleep (standing in for pack-objects) and prints its pid.
use tokio::io::AsyncReadExt;
let mut child = tokio::process::Command::new("sh")
.arg("-c")
.arg("sleep 300 & echo \"$!\"; wait")
.stdout(Stdio::piped())
.process_group(0)
.spawn()
.unwrap();
let pgid = child.id().map(|id| id as i32);
// Read the backgrounded grandchild's pid from the first stdout line.
let mut stdout = child.stdout.take().unwrap();
let mut buf = Vec::new();
loop {
let mut byte = [0u8; 1];
let n = stdout.read(&mut byte).await.unwrap();
if n == 0 || byte[0] == b'\n' {
break;
}
buf.push(byte[0]);
}
let grandchild: i32 = String::from_utf8(buf).unwrap().trim().parse().unwrap();
assert!(alive(grandchild), "grandchild should be running");
{
let _guard = KillGroupOnDrop { pgid };
} // group SIGTERM reaches sh AND the sleep grandchild
let _ = child.wait().await; // reap sh
// The grandchild reparents to init and is reaped; poll until it's gone.
let mut gone = false;
for _ in 0..200 {
if !alive(grandchild) {
gone = true;
break;
}
tokio::time::sleep(std::time::Duration::from_millis(10)).await;
}
assert!(
gone,
"grandchild must be terminated by the group signal (#53)"
);
}
#[cfg(unix)]
#[tokio::test]
async fn kill_group_guard_disarmed_does_not_kill() {
// A request that completed cleanly disarms the guard; dropping it must not
// signal anything.
let mut child = tokio::process::Command::new("sleep")
.arg("300")
.process_group(0)
.spawn()
.unwrap();
{
let mut guard = KillGroupOnDrop {
pgid: child.id().map(|id| id as i32),
};
guard.disarm();
} // disarmed -> no kill
assert!(
child.try_wait().unwrap().is_none(),
"disarmed guard must not kill the child"
);
// Clean up the still-running child.
let _ = child.kill().await;
let _ = child.wait().await;
}
}
+31 -7
View File
@@ -3,7 +3,9 @@
//! Repos are stored as `repos/v1/{owner_slug}/{repo_name}.tar.zst` — a
//! zstd-compressed tar archive of the bare repo directory.
use std::path::Path;
use std::collections::HashMap;
use std::path::{Path, PathBuf};
use std::sync::{Arc, Mutex, OnceLock};
use anyhow::{Context, Result};
use aws_sdk_s3::Client as S3Client;
@@ -110,8 +112,7 @@ impl TigrisClient {
.collect()
.await
.context("reading tigris response body")?
.into_bytes()
.to_vec();
.into_bytes();
// Extract tar.zst to local path
tokio::task::spawn_blocking({
@@ -156,6 +157,23 @@ fn compress_repo(repo_path: &Path) -> Result<Vec<u8>> {
Ok(compressed)
}
/// Per-repo-path lock serializing the publish (swap-into-place) step of
/// `decompress_repo`. Concurrent extractions unpack into isolated temp dirs in
/// parallel, but the final `remove_dir_all` + `rename` must not interleave for
/// the same `local_path`, or they race to a nondeterministic overwrite/failure.
fn publish_lock(local_path: &Path) -> Arc<Mutex<()>> {
// KNOWN LIMITATION: this map is never evicted — one (PathBuf, Arc<Mutex>)
// entry accrues per distinct repo path for the process lifetime. Bounded by
// the number of repos a node hosts, so it's negligible for normal use, but
// high-volume/churning deployments may want LRU or weak-ref eviction here.
static LOCKS: OnceLock<Mutex<HashMap<PathBuf, Arc<Mutex<()>>>>> = OnceLock::new();
let locks = LOCKS.get_or_init(|| Mutex::new(HashMap::new()));
let mut map = locks.lock().expect("publish lock map poisoned");
map.entry(local_path.to_path_buf())
.or_insert_with(|| Arc::new(Mutex::new(())))
.clone()
}
/// Decompress a tar.zst byte vector into a local directory.
///
/// Extraction is atomic with respect to `local_path`: the archive is unpacked
@@ -171,10 +189,12 @@ fn decompress_repo(data: &[u8], local_path: &Path) -> Result<()> {
.file_name()
.context("repo path has no file name")?
.to_string_lossy();
let tmp_dir = parent.join(format!(".{file_name}.tmp-extract"));
// Unique per-extraction temp dir: a fixed name would let two concurrent
// extractions of the same repo share one dir and clobber each other's
// in-progress unpack. A fresh UUID also means it can't collide with a
// leftover dir from a previously-interrupted run.
let tmp_dir = parent.join(format!(".{file_name}.tmp-extract.{}", uuid::Uuid::new_v4()));
// Clear any leftover temp dir from a previously-interrupted extraction.
let _ = std::fs::remove_dir_all(&tmp_dir);
std::fs::create_dir_all(&tmp_dir).context("creating temp extract dir")?;
// Unpack into the temp dir; on any failure, clean up and bail without
@@ -192,7 +212,11 @@ fn decompress_repo(data: &[u8], local_path: &Path) -> Result<()> {
// Swap the freshly-extracted repo into place. rename within the same parent
// is effectively atomic, but most platforms refuse to rename onto a
// non-empty dir, so remove the old copy first.
// non-empty dir, so remove the old copy first. Serialize this per repo path:
// concurrent extractions unpack into isolated temp dirs, but their swaps
// must not interleave or they race to a nondeterministic overwrite/failure.
let lock = publish_lock(local_path);
let _publish = lock.lock().expect("publish lock poisoned");
if local_path.exists() {
std::fs::remove_dir_all(local_path).context("removing stale repo dir")?;
}
File diff suppressed because it is too large. Load diff
+181
View File
@@ -3,11 +3,21 @@ use chrono::Utc;
use std::sync::Arc;
use uuid::Uuid;
use crate::auth::AuthenticatedDid;
use crate::db::{AgentTask, Db};
use crate::state::TaskEventBroadcast;
use super::types::{AgentTaskType, CreateTaskInput, FinishTaskInput};
/// The verified signer DID for this request, or an auth error (N2). The DID is
/// attached request-scoped by the `/graphql` `optional_signature` layer; its
/// absence means the request was unsigned, so every mutation rejects.
fn require_signer<'a>(ctx: &'a Context<'_>) -> Result<&'a str> {
ctx.data::<AuthenticatedDid>()
.map(|d| d.0.as_str())
.map_err(|_| async_graphql::Error::new("authentication required"))
}
pub struct MutationRoot;
#[Object]
@@ -18,6 +28,13 @@ impl MutationRoot {
delegator_did: String,
input: CreateTaskInput,
) -> Result<AgentTaskType> {
let caller = require_signer(ctx)?;
if !crate::api::did_matches(caller, &delegator_did) {
return Err(async_graphql::Error::new(
"delegator_did must be the authenticated signer",
));
}
let delegator_did = caller.to_string();
let db = ctx.data_unchecked::<Arc<Db>>();
let now = Utc::now().to_rfc3339();
let task = AgentTask {
@@ -47,6 +64,13 @@ impl MutationRoot {
id: String,
assignee_did: String,
) -> Result<AgentTaskType> {
let caller = require_signer(ctx)?;
if !crate::api::did_matches(caller, &assignee_did) {
return Err(async_graphql::Error::new(
"assignee_did must be the authenticated signer",
));
}
let assignee_did = caller.to_string();
let db = ctx.data_unchecked::<Arc<Db>>();
let tx = ctx.data_unchecked::<tokio::sync::broadcast::Sender<TaskEventBroadcast>>();
let task = db
@@ -70,8 +94,27 @@ impl MutationRoot {
by_did: String,
input: FinishTaskInput,
) -> Result<AgentTaskType> {
let caller = require_signer(ctx)?;
if !crate::api::did_matches(caller, &by_did) {
return Err(async_graphql::Error::new(
"by_did must be the authenticated signer",
));
}
let by_did = caller.to_string();
let db = ctx.data_unchecked::<Arc<Db>>();
let tx = ctx.data_unchecked::<tokio::sync::broadcast::Sender<TaskEventBroadcast>>();
// Authorize the actor: binding by_did to the signer is necessary but not
// sufficient — only the task's assignee may finish it.
let existing = db
.get_task(&id)
.await
.map_err(|e| async_graphql::Error::new(e.to_string()))?
.ok_or_else(|| async_graphql::Error::new("task not found"))?;
if !crate::api::did_matches(caller, existing.assignee_did.as_deref().unwrap_or_default()) {
return Err(async_graphql::Error::new(
"only the task assignee can complete it",
));
}
let task = db
.finish_task(&id, "completed", input.result.as_deref())
.await
@@ -93,8 +136,26 @@ impl MutationRoot {
by_did: String,
input: FinishTaskInput,
) -> Result<AgentTaskType> {
let caller = require_signer(ctx)?;
if !crate::api::did_matches(caller, &by_did) {
return Err(async_graphql::Error::new(
"by_did must be the authenticated signer",
));
}
let by_did = caller.to_string();
let db = ctx.data_unchecked::<Arc<Db>>();
let tx = ctx.data_unchecked::<tokio::sync::broadcast::Sender<TaskEventBroadcast>>();
// Authorize the actor: only the task's assignee may fail it.
let existing = db
.get_task(&id)
.await
.map_err(|e| async_graphql::Error::new(e.to_string()))?
.ok_or_else(|| async_graphql::Error::new("task not found"))?;
if !crate::api::did_matches(caller, existing.assignee_did.as_deref().unwrap_or_default()) {
return Err(async_graphql::Error::new(
"only the task assignee can fail it",
));
}
let reason = input.reason.unwrap_or_default();
let task = db
.finish_task(&id, "failed", Some(&reason))
@@ -110,3 +171,123 @@ impl MutationRoot {
Ok(AgentTaskType::from(task))
}
}
#[cfg(test)]
mod tests {
use crate::auth::AuthenticatedDid;
use async_graphql::{Request, Response};
use sqlx::PgPool;
fn errors(resp: &Response) -> String {
resp.errors
.iter()
.map(|e| e.message.clone())
.collect::<Vec<_>>()
.join("; ")
}
/// N2: GraphQL mutations require a verified signer and bind the acting DID to
/// it. Unsigned is rejected; a signer other than the claimed actor is
/// rejected; the matching signer passes the auth gate.
#[sqlx::test]
async fn mutation_requires_and_binds_signer(pool: PgPool) {
let state = crate::test_support::test_state(pool).await;
let schema = state.graphql_schema.as_ref();
let assignee = "did:key:zASSIGNEEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA";
let q = format!(
r#"mutation {{ claimTask(id: "no-such-task", assigneeDid: "{assignee}") {{ id }} }}"#
);
// 1. Unsigned → rejected before any DB work.
let resp = schema.execute(Request::new(&q)).await;
assert!(
errors(&resp).contains("authentication required"),
"unsigned mutation must be rejected: {}",
errors(&resp)
);
// 2. Signed as someone other than the claimed assignee → rejected.
let resp = schema
.execute(Request::new(&q).data(AuthenticatedDid(
"did:key:zOTHERBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB".into(),
)))
.await;
assert!(
errors(&resp).contains("authenticated signer"),
"DID mismatch must be rejected: {}",
errors(&resp)
);
// 3. Signed as the claimed assignee → passes the auth gate (any remaining
// error is the missing task, not an auth error).
let resp = schema
.execute(Request::new(&q).data(AuthenticatedDid(assignee.into())))
.await;
let errs = errors(&resp);
assert!(
!errs.contains("authentication required") && !errs.contains("authenticated signer"),
"matching signer must pass the auth gate: {errs}"
);
}
/// Adversarial-review GATE-1 (GraphQL): completing a task requires being its
/// assignee, not merely signing as the by_did you pass. A signer who is not
/// the assignee is rejected even though the by_did binding passes; the
/// assignee completes.
#[sqlx::test]
async fn complete_task_requires_assignee(pool: PgPool) {
let state = crate::test_support::test_state(pool).await;
let assignee = "did:key:zGQLASSIGNEEAAAAAAAAAAAAAAAAAAAAAAAAAAAA";
let stranger = "did:key:zGQLSTRANGERBBBBBBBBBBBBBBBBBBBBBBBBBBBB";
let now = chrono::Utc::now().to_rfc3339();
let task = crate::db::AgentTask {
id: "task-g".into(),
repo_id: None,
kind: "build".into(),
status: "pending".into(),
delegator_did: "did:key:zGQLDELEGATORCCCCCCCCCCCCCCCCCCCCCCCCCC".into(),
assignee_did: None,
capability: "repo:write".into(),
ucan_token: None,
payload: None,
result: None,
created_at: now.clone(),
updated_at: now,
deadline: None,
};
state.db.create_task(&task).await.expect("seed task");
state
.db
.claim_task("task-g", assignee)
.await
.expect("claim");
let schema = state.graphql_schema.as_ref();
let q = |actor: &str| {
format!(
r#"mutation {{ completeTask(id: "task-g", byDid: "{actor}", input: {{}}) {{ id status }} }}"#
)
};
// Stranger signs as themselves and passes byDid=self (so the signer
// binding passes), but is not the assignee → rejected by authorization.
let resp = schema
.execute(Request::new(q(stranger)).data(AuthenticatedDid(stranger.into())))
.await;
assert!(
errors(&resp).contains("assignee"),
"a non-assignee signer must be rejected: {}",
errors(&resp)
);
// The assignee completes with no error.
let resp = schema
.execute(Request::new(q(assignee)).data(AuthenticatedDid(assignee.into())))
.await;
assert!(
errors(&resp).is_empty(),
"the assignee should complete the task: {}",
errors(&resp)
);
}
}
+19 -1
View File
@@ -12,11 +12,29 @@ impl QueryRoot {
async fn repos(&self, ctx: &Context<'_>) -> Result<Vec<RepoType>> {
let db = ctx.data_unchecked::<Arc<Db>>();
let repos = db
.list_all_repos()
.list_all_repos_deduped()
.await
.map_err(|e| async_graphql::Error::new(e.to_string()))?;
// Apply the same "/" visibility gate the REST/per-repo endpoints use so
// this surface does not enumerate private repos (#97). The caller DID is
// threaded onto the context by optional_signature; absent = anonymous.
let caller = ctx
.data::<crate::auth::AuthenticatedDid>()
.ok()
.map(|d| d.0.as_str());
let ids: Vec<String> = repos.iter().map(|r| r.id.clone()).collect();
let rules_by_repo = db
.list_visibility_rules_for_repos(&ids)
.await
.map_err(|e| async_graphql::Error::new(e.to_string()))?;
Ok(repos
.into_iter()
.filter(|r| {
let rules = rules_by_repo.get(&r.id).map(Vec::as_slice).unwrap_or(&[]);
crate::visibility::listable_at_root(rules, r.is_public, &r.owner_did, caller)
})
.map(|r| RepoType {
name: r.name,
owner_did: r.owner_did,
+633
View File
@@ -0,0 +1,633 @@
//! iCaptcha proof-of-intelligence gate.
//!
//! Spam-prone endpoints (repo creation, agent registration) can require the
//! caller to present an iCaptcha proof: a small Ed25519-signed token minted by
//! <https://icaptcha.gitlawb.com> after the caller solves an escalating
//! challenge. We verify the proof OFFLINE (no per-request call to iCaptcha)
//! using its published public key, and bind each proof to the authenticated
//! agent DID so a proof cannot be shared between identities.
//!
//! Behaviour is controlled by `ICAPTCHA_MODE`:
//! * `off` (default) — gate is inert, nothing is checked.
//! * `shadow` — verify and log would-be rejections, but always allow.
//! * `enforce` — reject requests without a valid, sufficiently-strong proof.
//!
//! Config (env):
//! ICAPTCHA_MODE off | shadow | enforce (default off)
//! ICAPTCHA_URL base URL (default https://icaptcha.gitlawb.com)
//! ICAPTCHA_PUBKEY base64url Ed25519 public key (optional; else fetched from /v1/pubkey)
//! ICAPTCHA_REQUIRED_LEVEL minimum proof level (default 3)
use std::sync::OnceLock;
use std::time::{Duration, SystemTime, UNIX_EPOCH};
use axum::http::HeaderMap;
use base64::engine::general_purpose::URL_SAFE_NO_PAD;
use base64::Engine;
use ed25519_dalek::{Signature, VerifyingKey};
use serde::Deserialize;
use crate::error::AppError;
const PROOF_HEADER: &str = "x-icaptcha-proof";
/// How long a mirror-admission `jti` stays in the replay ledger. The direct
/// request path spends a proof against its own `exp` (it can't be presented once
/// expired). The propagation path deliberately accepts already-expired proofs,
/// so spending against `exp` would store an already-past expiry that the next
/// sweep deletes — letting the same token admit another mirror minutes later.
/// We instead retain the mirror replay record for a long, fixed window so the
/// per-node single-use property is durable.
const MIRROR_REPLAY_RETENTION_SECS: i64 = 365 * 24 * 60 * 60;
#[derive(Clone, Copy, PartialEq, Eq, Debug)]
pub enum Mode {
Off,
Shadow,
Enforce,
}
impl Mode {
fn as_str(self) -> &'static str {
match self {
Mode::Off => "off",
Mode::Shadow => "shadow",
Mode::Enforce => "enforce",
}
}
}
/// Parse `ICAPTCHA_MODE`. Returns `None` for unrecognized values so the caller
/// can surface the typo instead of silently disabling the gate.
fn parse_mode(s: &str) -> Option<Mode> {
match s.trim().to_ascii_lowercase().as_str() {
"" | "off" => Some(Mode::Off),
"shadow" => Some(Mode::Shadow),
"enforce" => Some(Mode::Enforce),
_ => None,
}
}
/// Parse `ICAPTCHA_REQUIRED_LEVEL`. Defaults to 3; warns (rather than silently
/// lowering the threshold) when a non-empty value fails to parse.
fn parse_required_level() -> u32 {
const DEFAULT: u32 = 3;
match std::env::var("ICAPTCHA_REQUIRED_LEVEL") {
Ok(v) if !v.trim().is_empty() => v.trim().parse().unwrap_or_else(|_| {
tracing::warn!(
value = %v,
default = DEFAULT,
"invalid ICAPTCHA_REQUIRED_LEVEL; using default"
);
DEFAULT
}),
_ => DEFAULT,
}
}
struct Verifier {
mode: Mode,
url: String,
required_level: u32,
key: Option<VerifyingKey>,
}
static VERIFIER: OnceLock<Verifier> = OnceLock::new();
#[derive(Deserialize, Debug)]
struct ProofClaims {
sub: String,
level: u32,
exp: i64,
/// Unique proof id, consumed once so a proof cannot be replayed.
jti: String,
}
#[derive(Deserialize)]
struct Jwk {
x: String,
}
#[derive(Deserialize)]
struct Jwks {
keys: Vec<Jwk>,
}
fn now_secs() -> i64 {
SystemTime::now()
.duration_since(UNIX_EPOCH)
.map(|d| d.as_secs() as i64)
.unwrap_or(0)
}
fn decode_key(b64url: &str) -> Option<VerifyingKey> {
let bytes = URL_SAFE_NO_PAD.decode(b64url.trim()).ok()?;
let arr: [u8; 32] = bytes.try_into().ok()?;
VerifyingKey::from_bytes(&arr).ok()
}
async fn fetch_key(url: &str) -> Option<VerifyingKey> {
let endpoint = format!("{}/v1/pubkey", url.trim_end_matches('/'));
// Bounded request: a hung /v1/pubkey must never block node startup. On
// timeout/error we return None and the gate stays inert (fail safe).
let client = reqwest::Client::builder()
.timeout(Duration::from_secs(5))
.build()
.ok()?;
let jwks: Jwks = client.get(&endpoint).send().await.ok()?.json().await.ok()?;
decode_key(&jwks.keys.first()?.x)
}
/// Initialize the gate from the environment. Call once at startup. Never panics;
/// if the gate is active but no key can be loaded, it stays inert and warns.
pub async fn init() {
let raw_mode = std::env::var("ICAPTCHA_MODE").unwrap_or_default();
let mode = parse_mode(&raw_mode).unwrap_or_else(|| {
tracing::warn!(value = %raw_mode, "invalid ICAPTCHA_MODE; disabling iCaptcha gate");
Mode::Off
});
let url = std::env::var("ICAPTCHA_URL")
.unwrap_or_else(|_| "https://icaptcha.gitlawb.com".to_string());
let required_level = parse_required_level();
let key = if mode == Mode::Off {
None
} else {
match std::env::var("ICAPTCHA_PUBKEY") {
Ok(b64) if !b64.is_empty() => decode_key(&b64),
_ => fetch_key(&url).await,
}
};
if mode != Mode::Off {
if key.is_some() {
tracing::info!(mode = mode.as_str(), required_level, "iCaptcha gate active");
} else {
tracing::warn!(
mode = mode.as_str(),
"iCaptcha gate enabled but no public key could be loaded; staying inert"
);
}
}
let _ = VERIFIER.set(Verifier {
mode,
url,
required_level,
key,
});
}
/// Outcome of the synchronous, IO-free decision step.
#[derive(Debug)]
enum Decision {
/// Allow the request (off, shadow, inert/no-key, or verified non-enforcing).
Allow,
/// Enforce mode and verification failed; reject with this reason.
Reject(String),
/// Enforce mode and the proof verified; the caller must consume its `jti`
/// (and reject replays) before allowing, then may persist it.
Consume(VerifiedProof),
}
/// A verified proof: the raw token plus the claims we act on. Carried by a
/// [`ProofGuard`] so that, once consumed, the gated handler can persist it with
/// the created repo — letting the proof travel with the repo when it propagates
/// to peers (see [`admit_mirror`]).
#[derive(Debug)]
pub struct VerifiedProof {
token: String,
sub: String,
level: u32,
jti: String,
exp: i64,
}
impl VerifiedProof {
/// Persist this proof against a freshly-created repo so a mirroring peer can
/// re-verify it. Best-effort callers may ignore failures; the gate's
/// security does not depend on it (propagation just falls back to quarantine).
pub async fn record_for_repo(&self, db: &crate::db::Db, repo_id: &str) -> Result<(), AppError> {
db.record_repo_proof(
repo_id,
&self.token,
&self.sub,
self.level as i32,
&self.jti,
self.exp,
)
.await?;
Ok(())
}
}
/// A verified proof awaiting consumption. Verification (which rejects invalid or
/// missing proofs) is separated from consumption (which spends the single-use
/// `jti`) so a request rejected by later validation never burns a valid proof.
/// The caller must `consume()` this guard immediately before the gated write.
/// For off/shadow/inert there is nothing to consume.
#[must_use = "a verified iCaptcha proof must be consumed before the gated action"]
pub struct ProofGuard(Option<VerifiedProof>);
impl ProofGuard {
/// Spend the proof's `jti` (single-use) and return the verified proof so the
/// caller can persist it. A replay is rejected. Returns `None` when there is
/// nothing to consume (off/shadow/inert).
pub async fn consume(self, db: &crate::db::Db) -> Result<Option<VerifiedProof>, AppError> {
match self.0 {
Some(p) => {
if !db.consume_proof_jti(&p.jti, p.exp).await? {
return Err(AppError::IcaptchaProofRequired(
"iCaptcha proof already used (replay); solve a fresh challenge".to_string(),
));
}
Ok(Some(p))
}
None => Ok(None),
}
}
}
/// Verify the proof in `headers` belongs to `did`. Rejects missing/invalid
/// proofs early (enforce mode); off/shadow never reject. Returns a [`ProofGuard`]
/// the caller must `consume()` right before the gated write. Off/shadow/inert
/// yield a no-op guard that consumes nothing.
pub fn verify_request(headers: &HeaderMap, did: &str) -> Result<ProofGuard, AppError> {
let v = match VERIFIER.get() {
Some(v) => v,
None => return Ok(ProofGuard(None)), // not initialized -> inert
};
match decide(v, headers, did, now_secs()) {
Decision::Allow => Ok(ProofGuard(None)),
Decision::Reject(reason) => Err(reject_error(v, &reason)),
Decision::Consume(proof) => Ok(ProofGuard(Some(proof))),
}
}
fn reject_error(v: &Verifier, reason: &str) -> AppError {
AppError::IcaptchaProofRequired(format!(
"iCaptcha proof required ({reason}). Solve a challenge at {} for level >= {} and resend with the {} header.",
v.url, v.required_level, PROOF_HEADER
))
}
/// Mode-aware decision. Pure and IO-free (no DB; clock injected via `now`) so it
/// is fully unit-testable. The caller performs jti consumption for `Consume`.
fn decide(v: &Verifier, headers: &HeaderMap, did: &str, now: i64) -> Decision {
if v.mode == Mode::Off {
return Decision::Allow;
}
// Fail safe: if no public key could be loaded (e.g. iCaptcha was unreachable
// at startup), stay inert rather than rejecting every request. The operator
// already saw a startup warning. An iCaptcha hiccup must never break repo
// creation or registration.
if v.key.is_none() {
return Decision::Allow;
}
let token = headers.get(PROOF_HEADER).and_then(|h| h.to_str().ok());
let result = match token {
Some(t) => verify_token(v, t, did, now, true).map(|claims| (t, claims)),
None => Err("missing proof header".to_string()),
};
match result {
Ok((token, claims)) => match v.mode {
Mode::Enforce => Decision::Consume(VerifiedProof {
token: token.to_string(),
sub: claims.sub,
level: claims.level,
jti: claims.jti,
exp: claims.exp,
}),
// Shadow/Off: never reject, and do not consume (observational only).
_ => Decision::Allow,
},
Err(reason) => match v.mode {
Mode::Shadow => {
tracing::warn!(did = %did, reason, "iCaptcha (shadow) would reject");
Decision::Allow
}
Mode::Enforce => Decision::Reject(reason),
Mode::Off => Decision::Allow,
},
}
}
/// Header-extracting wrapper over [`verify_token`] (enforces expiry). The
/// production path inlines this in [`decide`]; retained as the unit-test entry
/// point that also exercises header extraction.
#[cfg(test)]
fn verify(v: &Verifier, headers: &HeaderMap, did: &str, now: i64) -> Result<ProofClaims, String> {
let proof = headers
.get(PROOF_HEADER)
.and_then(|h| h.to_str().ok())
.ok_or("missing proof header")?;
verify_token(v, proof, did, now, true)
}
/// Core verification of a raw proof token, separated for testability and reuse.
/// Checks the Ed25519 signature, the required level, and the DID binding. `exp`
/// is only enforced when `check_exp` is true: the direct request path enforces
/// freshness, but the propagation path ([`admit_mirror`]) does not — a proof has
/// usually expired by the time a repo mirrors, and the origin enforced freshness
/// at creation. `now` is unix seconds.
fn verify_token(
v: &Verifier,
token: &str,
did: &str,
now: i64,
check_exp: bool,
) -> Result<ProofClaims, String> {
let key = v.key.as_ref().ok_or("verifier has no public key")?;
let (payload, sig_b64) = token.split_once('.').ok_or("malformed proof")?;
let sig_bytes = URL_SAFE_NO_PAD
.decode(sig_b64)
.map_err(|_| "bad signature encoding")?;
let sig = Signature::from_slice(&sig_bytes).map_err(|_| "bad signature length")?;
key.verify_strict(payload.as_bytes(), &sig)
.map_err(|_| "signature verification failed")?;
let claims_bytes = URL_SAFE_NO_PAD
.decode(payload)
.map_err(|_| "bad payload encoding")?;
let claims: ProofClaims = serde_json::from_slice(&claims_bytes).map_err(|_| "bad claims")?;
if check_exp && claims.exp < now {
return Err("proof expired".to_string());
}
if claims.level < v.required_level {
return Err(format!(
"level {} below required {}",
claims.level, v.required_level
));
}
if !crate::api::did_matches(did, &claims.sub) {
return Err("proof subject does not match authenticated DID".to_string());
}
Ok(claims)
}
/// Outcome of admitting a repo mirrored from a peer.
#[derive(Debug, PartialEq, Eq)]
pub enum MirrorAdmission {
/// Admit the mirror normally (off/shadow/inert, or a valid proof in enforce).
Admit,
/// Enforce mode and the mirror's proof is missing/invalid/replayed — mirror
/// it but quarantine until an operator releases it. Carries the reason.
Quarantine(String),
}
/// Decide whether to admit a repo being mirrored from a peer, given the proof
/// token the origin served (`None` if it had none or the fetch failed) and the
/// mirrored repo's owner DID.
///
/// Off/shadow/inert always admit (shadow logs the would-be quarantine and writes
/// nothing). In enforce mode the proof is re-verified offline against the local
/// public key — signature, level, and owner-DID binding, but NOT expiry — and
/// its `jti` is consumed in the local ledger, so one solved challenge admits at
/// most one mirror per node (a malicious origin cannot reuse one proof to bless
/// many spam repos here).
///
/// LIMITATION (proof is not repo-bound): an iCaptcha proof commits only to the
/// solver's DID (`sub`), never to a repo — the proof is minted before the repo
/// exists. So the guarantee enforced here is "a level≥N challenge was solved by
/// this repo's owner DID", not "the owner intended THIS repo". A malicious origin
/// that harvests a victim's public proof (served by `get_icaptcha_proof`) can use
/// it to get ONE spam repo nominally owned by that victim admitted per node — the
/// per-node `jti` single-use caps the amplification at one repo per distinct
/// harvested proof, but does not eliminate it. Fully binding a proof to a repo
/// would require the iCaptcha service to embed a repo/target identifier in the
/// signed challenge (a protocol change, tracked as future work).
pub async fn admit_mirror(
db: &crate::db::Db,
token: Option<&str>,
owner_did: &str,
) -> MirrorAdmission {
let v = match VERIFIER.get() {
Some(v) => v,
None => return MirrorAdmission::Admit, // not initialized -> inert
};
// Off, or inert because no key could be loaded: admit unconditionally.
if v.mode == Mode::Off || v.key.is_none() {
return MirrorAdmission::Admit;
}
let shadow = v.mode == Mode::Shadow;
// Quarantine in enforce; in shadow just log and admit (observational, no IO).
let quarantine = |reason: String| -> MirrorAdmission {
if shadow {
tracing::warn!(owner = %owner_did, reason, "iCaptcha (shadow) would quarantine mirror");
MirrorAdmission::Admit
} else {
MirrorAdmission::Quarantine(reason)
}
};
let token = match token {
Some(t) => t,
None => return quarantine("origin served no iCaptcha proof".to_string()),
};
match verify_token(v, token, owner_did, now_secs(), false) {
Ok(claims) => {
if shadow {
// Observational: a valid proof would admit. Do not consume, so a
// later switch to enforce still sees the jti as fresh.
return MirrorAdmission::Admit;
}
// Retain the replay record on a fixed forward window, NOT the proof's
// own (already-past) exp — otherwise the next sweep frees the jti and
// the same token could admit another mirror here.
let retain_until = now_secs().saturating_add(MIRROR_REPLAY_RETENTION_SECS);
match db.consume_proof_jti(&claims.jti, retain_until).await {
Ok(true) => MirrorAdmission::Admit,
Ok(false) => MirrorAdmission::Quarantine(
"iCaptcha proof already used to admit another mirror".to_string(),
),
Err(e) => {
tracing::warn!(owner = %owner_did, err = %e, "iCaptcha mirror ledger error; quarantining");
MirrorAdmission::Quarantine("iCaptcha proof ledger unavailable".to_string())
}
}
}
Err(reason) => quarantine(reason),
}
}
#[cfg(test)]
mod tests {
use super::*;
// Real values captured from https://icaptcha.gitlawb.com (a live proof).
const PUBKEY_X: &str = "xjyPNqIbvc9U-kwXW6u9mDqRJ7E2UUMOaJdUWhpEXq8";
const PROOF: &str = "eyJzdWIiOiJkaWQ6a2V5Onp0ZXN0IiwibGV2ZWwiOjMsImlzcyI6ImljYXB0Y2hhIiwiaWF0IjoxNzgyNTcyODUxLCJleHAiOjE3ODI1NzMxNTEsImp0aSI6IjRiNTIyOGE1YmVkNzEyMmRlZTlmNDdmZiJ9.5UXVPZ8Eo91VnlcvgDXtW-Fx7J2jr7h535SAstQEpigxBr7FF7V6R0XB4PBDgdoBPnhdH_kVEfRPfdHPSdB0CA";
const SUB: &str = "did:key:ztest";
const IAT: i64 = 1782572851; // within the proof's validity window
fn verifier(level: u32) -> Verifier {
Verifier {
mode: Mode::Enforce,
url: "https://icaptcha.gitlawb.com".to_string(),
required_level: level,
key: decode_key(PUBKEY_X),
}
}
fn headers_with(proof: &str) -> HeaderMap {
let mut h = HeaderMap::new();
h.insert(PROOF_HEADER, proof.parse().unwrap());
h
}
#[test]
fn accepts_a_real_proof() {
let v = verifier(3);
assert!(verify(&v, &headers_with(PROOF), SUB, IAT).is_ok());
}
#[test]
fn rejects_expired_proof() {
let v = verifier(3);
let err = verify(&v, &headers_with(PROOF), SUB, 9_999_999_999).unwrap_err();
assert!(err.contains("expired"), "{err}");
}
#[test]
fn rejects_wrong_did() {
let v = verifier(3);
let err = verify(&v, &headers_with(PROOF), "did:key:zsomeoneelse", IAT).unwrap_err();
assert!(err.contains("subject"), "{err}");
}
#[test]
fn rejects_insufficient_level() {
let v = verifier(5); // proof is level 3
let err = verify(&v, &headers_with(PROOF), SUB, IAT).unwrap_err();
assert!(err.contains("below required"), "{err}");
}
#[test]
fn rejects_tampered_signature() {
let v = verifier(3);
// Flip one base64url char in the signature so it is guaranteed different.
let (payload, sig) = PROOF.split_once('.').unwrap();
let mut chars: Vec<char> = sig.chars().collect();
chars[0] = if chars[0] == 'A' { 'B' } else { 'A' };
let tampered = format!("{}.{}", payload, chars.into_iter().collect::<String>());
assert!(verify(&v, &headers_with(&tampered), SUB, IAT).is_err());
}
#[test]
fn rejects_missing_header() {
let v = verifier(3);
let err = verify(&v, &HeaderMap::new(), SUB, IAT).unwrap_err();
assert!(err.contains("missing"), "{err}");
}
#[test]
fn parse_mode_accepts_documented_values_and_rejects_junk() {
assert_eq!(parse_mode(""), Some(Mode::Off));
assert_eq!(parse_mode("off"), Some(Mode::Off));
assert_eq!(parse_mode(" Shadow "), Some(Mode::Shadow));
assert_eq!(parse_mode("ENFORCE"), Some(Mode::Enforce));
// Typos must NOT silently disable the gate.
assert_eq!(parse_mode("enforced"), None);
assert_eq!(parse_mode("on"), None);
}
#[test]
fn off_mode_allows_everything() {
let mut v = verifier(3);
v.mode = Mode::Off;
assert!(matches!(
decide(&v, &HeaderMap::new(), SUB, IAT),
Decision::Allow
));
}
#[test]
fn enforce_without_key_stays_inert() {
// iCaptcha unreachable at startup -> no key -> must not reject.
let v = Verifier {
mode: Mode::Enforce,
url: "https://icaptcha.gitlawb.com".to_string(),
required_level: 3,
key: None,
};
assert!(matches!(
decide(&v, &HeaderMap::new(), SUB, IAT),
Decision::Allow
));
}
#[test]
fn enforce_with_key_rejects_missing_proof() {
let v = verifier(3);
assert!(matches!(
decide(&v, &HeaderMap::new(), SUB, IAT),
Decision::Reject(_)
));
}
#[test]
fn shadow_allows_despite_bad_proof() {
let mut v = verifier(3);
v.mode = Mode::Shadow;
assert!(matches!(
decide(&v, &HeaderMap::new(), SUB, IAT),
Decision::Allow
));
}
#[test]
fn enforce_valid_proof_requires_consuming_its_jti() {
// A verified proof under enforce must yield Consume carrying the jti, so
// the caller can spend it once and reject replays.
let v = verifier(3);
match decide(&v, &headers_with(PROOF), SUB, IAT) {
Decision::Consume(p) => {
assert_eq!(p.jti, "4b5228a5bed7122dee9f47ff");
assert_eq!(p.exp, 1782573151);
assert_eq!(p.token, PROOF, "the raw token is retained for persistence");
assert_eq!(p.sub, SUB);
}
other => panic!("expected Consume, got {other:?}"),
}
}
#[test]
fn propagation_accepts_an_expired_proof() {
// The mirror-admission path verifies sig/level/DID but NOT expiry, since a
// proof has usually expired by the time a repo propagates. The same proof
// that the direct path rejects as expired must verify here.
let v = verifier(3);
let now = 9_999_999_999; // far past the proof's exp
assert!(
verify_token(&v, PROOF, SUB, now, true).is_err(),
"direct path (check_exp=true) rejects the expired proof"
);
assert!(
verify_token(&v, PROOF, SUB, now, false).is_ok(),
"propagation path (check_exp=false) accepts it"
);
}
#[test]
fn propagation_still_enforces_signature_level_and_did() {
let v = verifier(3);
// wrong owner DID
assert!(verify_token(&v, PROOF, "did:key:zother", IAT, false).is_err());
// insufficient level
let v5 = verifier(5);
assert!(verify_token(&v5, PROOF, SUB, IAT, false).is_err());
// tampered signature
let (payload, sig) = PROOF.split_once('.').unwrap();
let mut chars: Vec<char> = sig.chars().collect();
chars[0] = if chars[0] == 'A' { 'B' } else { 'A' };
let tampered = format!("{}.{}", payload, chars.into_iter().collect::<String>());
assert!(verify_token(&v, &tampered, SUB, IAT, false).is_err());
}
}
+10 -44
View File
@@ -7,8 +7,6 @@
//! If `ipfs_api` is empty the functions are no-ops, so the node works fine
//! without a local IPFS daemon.
use std::collections::HashSet;
use anyhow::Result;
use gitlawb_core::cid::Cid;
@@ -85,31 +83,27 @@ pub async fn cat(ipfs_api: &str, cid: &str) -> Result<Vec<u8>> {
Ok(resp.bytes().await?.to_vec())
}
/// List all git objects in the given bare repo and pin any that are not yet
/// recorded in `pinned_cids`.
/// Pin any of the given candidate git objects that are not yet recorded in
/// `pinned_cids`.
///
/// `object_list` is the already-withheld-filtered OID set to pin: the caller
/// applies `visibility_pack::replicable_objects` on the delta path or the
/// `..._fail_closed` filter on the full-scan path before calling, so this
/// function never sees a withheld blob. `repo_path` is still needed to read each
/// object's bytes. The twin in `pinata.rs` mirrors this shape — change both in
/// lockstep.
///
/// Returns a list of `(sha256_hex, cid)` pairs for objects pinned this call.
pub async fn pin_new_objects(
ipfs_api: &str,
repo_path: &std::path::Path,
object_list: Vec<String>,
db: &crate::db::Db,
withheld: &HashSet<String>,
) -> Vec<(String, String)> {
if ipfs_api.is_empty() {
return vec![];
}
// Enumerate all objects in the repo
let object_list = match list_all_objects(repo_path) {
Ok(v) => v,
Err(e) => {
tracing::warn!(repo = %repo_path.display(), err = %e, "failed to list git objects for IPFS pinning");
return vec![];
}
};
let object_list = crate::git::visibility_pack::replicable_objects(object_list, withheld);
let mut pinned = Vec::new();
for sha in object_list {
@@ -150,31 +144,3 @@ pub async fn pin_new_objects(
pinned
}
/// Run `git cat-file --batch-all-objects --batch-check='%(objectname)'`
/// to get all object SHA-256 hashes in the repository.
fn list_all_objects(repo_path: &std::path::Path) -> Result<Vec<String>> {
let output = std::process::Command::new("git")
.args([
"cat-file",
"--batch-all-objects",
"--batch-check=%(objectname)",
])
.current_dir(repo_path)
.output()
.map_err(|e| anyhow::anyhow!("failed to run git cat-file: {e}"))?;
if !output.status.success() {
let stderr = String::from_utf8_lossy(&output.stderr);
return Err(anyhow::anyhow!("git cat-file failed: {stderr}"));
}
let stdout = String::from_utf8_lossy(&output.stdout);
let hashes = stdout
.lines()
.map(|l| l.trim().to_string())
.filter(|l| !l.is_empty())
.collect();
Ok(hashes)
}
+29 -1
View File
@@ -9,6 +9,7 @@ mod encrypted_pin;
mod error;
mod git;
mod graphql;
mod icaptcha;
mod ipfs_pin;
mod metrics;
mod operator;
@@ -18,6 +19,8 @@ mod rate_limit;
mod server;
mod state;
mod sync;
#[cfg(test)]
mod test_support;
mod visibility;
mod webhooks;
@@ -98,6 +101,15 @@ async fn main() -> Result<()> {
}
}
// Prune peer rows with non-public hosts (loopback/private/internal) that
// were injected via the unauthenticated announce route — they poison the
// sync-notify fan-out (SSRF + crowding out real peers).
match db.prune_non_public_peers().await {
Ok(0) => {}
Ok(n) => info!(removed = n, "pruned non-public (poisoned) peer rows"),
Err(e) => warn!(err = %e, "prune_non_public_peers failed (non-fatal)"),
}
// Ensure repos directory exists
std::fs::create_dir_all(&config.repos_dir).context("failed to create repos directory")?;
@@ -133,9 +145,14 @@ async fn main() -> Result<()> {
None
};
// Do not follow redirects: this client fans out to peer-supplied URLs
// (sync trigger, profile/repo fetches). A peer answering `302 Location:
// http://127.0.0.1/` would otherwise bypass the announce-time public-URL
// check and turn the redirect into an SSRF.
let http_client = Arc::new(
reqwest::Client::builder()
.timeout(std::time::Duration::from_secs(10))
.redirect(reqwest::redirect::Policy::none())
.build()?,
);
@@ -175,6 +192,9 @@ async fn main() -> Result<()> {
let rate_limiter = rate_limit::RateLimiter::new(10, std::time::Duration::from_secs(3600));
// Initialize the iCaptcha proof gate (inert unless ICAPTCHA_MODE is set).
icaptcha::init().await;
let state = AppState {
config: Arc::new(config.clone()),
db,
@@ -245,15 +265,23 @@ async fn main() -> Result<()> {
});
}
// Periodic cleanup of expired rate limit entries
// Periodic cleanup of expired rate limit entries + consumed-proof ledger
{
let rl = state.rate_limiter.clone();
let db = state.db.clone();
let mut shutdown_rx = state.subscribe_shutdown();
tokio::spawn(async move {
loop {
tokio::select! {
_ = tokio::time::sleep(std::time::Duration::from_secs(300)) => {
rl.cleanup().await;
let now = std::time::SystemTime::now()
.duration_since(std::time::UNIX_EPOCH)
.map(|d| d.as_secs() as i64)
.unwrap_or(0);
if let Err(e) = db.sweep_expired_proofs(now).await {
tracing::warn!(err = %e, "failed to sweep expired iCaptcha proofs");
}
}
_ = shutdown_rx.changed() => {
if *shutdown_rx.borrow() {
+10 -41
View File
@@ -7,7 +7,6 @@
//! no-op, so nodes without Pinata backing work fine.
use anyhow::Result;
use std::collections::HashSet;
/// Pin a single git object's raw bytes on Pinata (v3 API).
///
@@ -67,35 +66,28 @@ pub async fn pin_object(
Ok(cid)
}
/// Pin all git objects in a repo that haven't yet been sent to Pinata.
/// Pin any of the given candidate git objects that haven't yet been sent to
/// Pinata.
///
/// Objects already recorded with a `pinata_cid` in the DB are skipped.
/// Returns `(sha_hex, cid)` pairs for each newly pinned object.
/// `object_list` is the already-withheld-filtered OID set to pin: the caller
/// applies `visibility_pack::replicable_objects` on the delta path or the
/// `..._fail_closed` filter on the full-scan path before calling. `repo_path` is
/// still needed to read each object's bytes. The twin in `ipfs_pin.rs` mirrors
/// this shape — change both in lockstep. Objects already recorded with a
/// `pinata_cid` are skipped. Returns `(sha_hex, cid)` pairs for each newly
/// pinned object.
pub async fn pin_new_objects(
client: &reqwest::Client,
upload_url: &str,
jwt: &str,
repo_path: &std::path::Path,
object_list: Vec<String>,
db: &crate::db::Db,
withheld: &HashSet<String>,
) -> Vec<(String, String)> {
if jwt.is_empty() {
return vec![];
}
let object_list = match list_all_objects(repo_path) {
Ok(v) => v,
Err(e) => {
tracing::warn!(
repo = %repo_path.display(),
err = %e,
"failed to list git objects for Pinata upload"
);
return vec![];
}
};
let object_list = crate::git::visibility_pack::replicable_objects(object_list, withheld);
let mut pinned = Vec::new();
for sha in object_list {
@@ -134,29 +126,6 @@ pub async fn pin_new_objects(
pinned
}
fn list_all_objects(repo_path: &std::path::Path) -> Result<Vec<String>> {
let out = std::process::Command::new("git")
.args([
"cat-file",
"--batch-all-objects",
"--batch-check=%(objectname)",
])
.current_dir(repo_path)
.output()
.map_err(|e| anyhow::anyhow!("failed to run git cat-file: {e}"))?;
if !out.status.success() {
let stderr = String::from_utf8_lossy(&out.stderr);
return Err(anyhow::anyhow!("git cat-file failed: {stderr}"));
}
Ok(String::from_utf8_lossy(&out.stdout)
.lines()
.map(|l| l.trim().to_string())
.filter(|l| !l.is_empty())
.collect())
}
// ── Tests ─────────────────────────────────────────────────────────────────────
#[cfg(test)]
+11 -3
View File
@@ -34,9 +34,17 @@ impl RateLimiter {
async fn check(&self, key: &str) -> bool {
let now = Instant::now();
let mut state = self.state.lock().await;
let window = state.entry(key.to_string()).or_insert_with(|| Window {
timestamps: Vec::new(),
});
// Look up before inserting so the common case (key already tracked)
// doesn't allocate a String per request.
if !state.contains_key(key) {
state.insert(
key.to_string(),
Window {
timestamps: Vec::new(),
},
);
}
let window = state.get_mut(key).expect("window just ensured");
window
.timestamps
.retain(|t| now.duration_since(*t) < self.window);
+56 -10
View File
@@ -20,8 +20,19 @@ use crate::auth;
use crate::rate_limit;
use crate::state::AppState;
async fn graphql_handler(State(state): State<AppState>, req: GraphQLRequest) -> GraphQLResponse {
state.graphql_schema.execute(req.into_inner()).await.into()
async fn graphql_handler(
State(state): State<AppState>,
auth: Option<axum::Extension<crate::auth::AuthenticatedDid>>,
req: GraphQLRequest,
) -> GraphQLResponse {
// `optional_signature` attaches the verified DID when a signature is present.
// Thread it into request-scoped GraphQL data; mutations enforce its presence
// in-resolver (N2) while queries stay open.
let mut inner = req.into_inner();
if let Some(axum::Extension(did)) = auth {
inner = inner.data(did);
}
state.graphql_schema.execute(inner).await.into()
}
async fn graphql_playground() -> impl IntoResponse {
@@ -49,6 +60,10 @@ pub fn build_router(state: AppState) -> Router {
let schema = state.graphql_schema.as_ref().clone();
let graphql_routes = Router::new()
.route("/graphql", get(graphql_playground).post(graphql_handler))
// Attach the verified DID to /graphql when a signature is present. The
// layer covers only routes added before it, so /graphql/ws (added after,
// read-only subscriptions) stays open.
.layer(middleware::from_fn(auth::optional_signature))
.route_service("/graphql/ws", GraphQLSubscription::new(schema));
// ── Task routes (write — require HTTP Signature) ───────────────────────
@@ -147,6 +162,10 @@ pub fn build_router(state: AppState) -> Router {
axum::routing::put(visibility::set_visibility)
.delete(visibility::remove_visibility)
.get(visibility::list_visibility),
)
.route(
"/api/v1/agents/{did}",
axum::routing::delete(agents::deregister_agent),
),
state.clone(),
);
@@ -168,9 +187,14 @@ pub fn build_router(state: AppState) -> Router {
);
// ── IPFS content-addressed retrieval and pin listing ──────────────────
// `/ipfs/{cid}` carries `optional_signature` so `get_by_cid` sees the caller
// identity and can apply per-repo visibility (#110); anonymous callers stay
// anonymous and still read genuinely public content. `/api/v1/ipfs/pins`
// stays unsigned — gating the pin index is tracked separately (#121).
let ipfs_routes = Router::new()
.route("/ipfs/{cid}", get(ipfs::get_by_cid))
.route("/api/v1/ipfs/pins", get(ipfs::list_pins));
.layer(middleware::from_fn(auth::optional_signature))
.merge(Router::new().route("/api/v1/ipfs/pins", get(ipfs::list_pins)));
// ── Arweave permanent anchors ──────────────────────────────────────────
let arweave_routes = Router::new().route("/api/v1/arweave/anchors", get(arweave::list_anchors));
@@ -357,6 +381,10 @@ pub fn build_router(state: AppState) -> Router {
"/api/v1/repos/{owner}/{repo}/withheld-paths",
axum::routing::get(visibility::withheld_paths),
)
.route(
"/api/v1/repos/{owner}/{repo}/icaptcha-proof",
axum::routing::get(repos::get_icaptcha_proof),
)
.route(
"/api/v1/repos/{owner}/{repo}/encrypted-blobs",
axum::routing::get(crate::api::encrypted::list_encrypted_blobs),
@@ -432,13 +460,31 @@ async fn node_info(State(state): State<AppState>) -> Json<serde_json::Value> {
}))
}
async fn stats(State(state): State<AppState>) -> Json<serde_json::Value> {
let repos = state
.db
.list_all_repos()
.await
.map(|r| r.len() as i64)
.unwrap_or(0);
pub(crate) async fn stats(State(state): State<AppState>) -> Json<serde_json::Value> {
// Count only the repos an anonymous caller could list, so the aggregate does
// not leak the existence of private/mode-A repos (#104 count oracle). Mirror
// the listing seam (api/repos.rs): over-fetch the deduped set, batch-load the
// visibility rules, and keep rows that pass listable_at_root. The caller is
// always None — meta_routes carries no auth layer (see the route group in this
// file). Fail closed: any DB error collapses the whole count to 0 (an
// under-count never leaks existence), preserving the prior `.unwrap_or(0)`.
let repos = async {
// stats only needs the count, so use the no-stars deduped list (same
// DEDUP_CTE) and skip the repo_stars aggregation the listing path needs.
let rows = state.db.list_all_repos_deduped().await?;
let ids: Vec<String> = rows.iter().map(|r| r.id.clone()).collect();
let rules_by_repo = state.db.list_visibility_rules_for_repos(&ids).await?;
let count = rows
.iter()
.filter(|r| {
let rules = rules_by_repo.get(&r.id).map(Vec::as_slice).unwrap_or(&[]);
crate::visibility::listable_at_root(rules, r.is_public, &r.owner_did, None)
})
.count() as i64;
Ok::<i64, anyhow::Error>(count)
}
.await
.unwrap_or(0);
let agents = state.db.count_agents().await.unwrap_or(0);
let pushes = state.db.count_pushes().await.unwrap_or(0);
Json(json!({
+310 -33
View File
@@ -32,19 +32,50 @@ enum MirrorMode {
Promisor,
}
/// Decide the mirror mode from the origin's `withheld-paths` response.
/// The on-disk promisor state of an existing mirror, read from
/// `remote.origin.promisor`. Three-valued so a git error is not mistaken for a
/// definitive "not a promisor": `git config --get` collapses "key absent" and
/// "git failed" into the same non-zero exit otherwise, and treating an errored
/// probe as `NotPromisor` would let a transient failure downgrade a still-withheld
/// mirror (issue #48).
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
enum PromisorProbe {
/// `remote.origin.promisor` is `"true"`.
Promisor,
/// The key is absent (git exit 1) or set to a non-`true` value.
NotPromisor,
/// The probe itself failed (git spawn error or other non-zero exit).
Unknown,
}
/// Decide the mirror mode from the origin's `withheld-paths` response plus, when
/// the response is unknown, the existing mirror's on-disk promisor state.
///
/// `Some(non-empty)` → the repo has a private subtree → `Promisor`.
/// `Some(empty)` → fully public → `Plain`.
/// `None` → the lookup 404'd or failed. Attempt a `Plain` mirror; a
/// mode-A repo also 404s the git read endpoint, so the clone
/// fails and nothing is mirrored (fail-closed at the git
/// layer), while a public repo on a peer that predates the
/// `withheld-paths` route still gets mirrored.
fn classify_mirror(withheld: Option<Vec<String>>) -> MirrorMode {
/// `None` → the lookup 404'd or failed; the answer is *unknown*, which
/// is not the same as "public". For a fresh clone this stays
/// `Plain` (a mode-A repo also 404s the git read endpoint, so
/// the clone fails and nothing is mirrored — fail-closed at the
/// git layer — while a public repo on a peer that predates the
/// `withheld-paths` route still gets mirrored). For an existing
/// mirror it *biases toward preserving* the promisor state: a
/// genuine public transition returns `Some(empty)`, so on
/// `None` we cannot distinguish "still withheld, unreachable"
/// from "newly public, unreachable" and prefer the recoverable
/// choice over destroying the partial-clone config. An
/// indeterminate probe (`Unknown`) preserves for the same
/// reason (defense-in-depth, #48).
fn resolve_mirror_mode(
withheld: Option<Vec<String>>,
exists: bool,
promisor: PromisorProbe,
) -> MirrorMode {
match withheld {
Some(globs) if !globs.is_empty() => MirrorMode::Promisor,
_ => MirrorMode::Plain,
Some(_) => MirrorMode::Plain,
None if exists && promisor != PromisorProbe::NotPromisor => MirrorMode::Promisor,
None => MirrorMode::Plain,
}
}
@@ -109,10 +140,17 @@ async fn run(
let machine_id = std::env::var("FLY_MACHINE_ID").ok();
// Bound each peer HTTP call (withheld-paths lookup + replica registration)
// so a stalled peer cannot hang the worker.
// No redirects: peer URLs are attacker-influenceable, so a 3xx to a
// loopback/private address must not be followed (SSRF guard, matching the
// shared http_client and announce-time validation).
// Panic rather than fall back to reqwest::Client::new(): the default
// builder follows redirects, which would silently reintroduce the SSRF
// vector Policy::none() is here to close.
let client = reqwest::Client::builder()
.timeout(std::time::Duration::from_secs(30))
.redirect(reqwest::redirect::Policy::none())
.build()
.unwrap_or_else(|_| reqwest::Client::new());
.expect("failed to build no-redirect sync HTTP client");
info!("sync worker started (auto_sync=true)");
let mut interval = tokio::time::interval(std::time::Duration::from_secs(30));
loop {
@@ -130,6 +168,17 @@ async fn run(
}
}
/// Find the origin node's base HTTP URL for a sync item, trimming any trailing
/// slash so callers can append `/{path}` cleanly. Returns `None` when no peer
/// row matches the item's origin DID. Kept pure (no DB) so the per-batch peer
/// resolution can be unit-tested without a database.
fn resolve_origin_url(peers: &[crate::db::PeerRecord], node_did: &str) -> Option<String> {
peers
.iter()
.find(|p| p.did == node_did)
.map(|p| p.http_url.trim_end_matches('/').to_string())
}
async fn process_batch(
db: &Db,
config: &Config,
@@ -145,19 +194,27 @@ async fn process_batch(
}
};
for item in items {
// Resolve origin node HTTP URL from peer table
let peers = match db.list_peers().await {
Ok(p) => p,
Err(e) => {
warn!(err = %e, "failed to list peers for sync");
let _ = db.mark_sync_failed(&item.id).await;
continue;
}
};
// Nothing queued — the common case on the 30s idle poll. Bail before the
// peer lookup so an empty batch costs zero extra DB round-trips.
if items.is_empty() {
return;
}
let origin_url = match peers.iter().find(|p| p.did == item.node_did) {
Some(p) => p.http_url.trim_end_matches('/').to_string(),
// Resolve the peer table once per batch — every item only needs a lookup.
let peers = match db.list_peers().await {
Ok(p) => p,
Err(e) => {
warn!(err = %e, "failed to list peers for sync");
for item in &items {
let _ = db.mark_sync_failed(&item.id).await;
}
return;
}
};
for item in items {
let origin_url = match resolve_origin_url(&peers, &item.node_did) {
Some(url) => url,
None => {
warn!(node_did = %item.node_did, repo = %item.repo, "no peer URL found for sync origin — skipping");
let _ = db.mark_sync_failed(&item.id).await;
@@ -186,9 +243,30 @@ async fn process_batch(
let remote_url = format!("{}/{}", origin_url, item.repo);
let withheld = fetch_withheld(client, &origin_url, owner_short, repo_name).await;
let mode = classify_mirror(withheld);
let exists = local_path.exists();
let lookup_unknown = withheld.is_none();
// Only probe the on-disk promisor state when the lookup is unknown and the
// repo already exists — the sole case where it changes the resolved mode.
let promisor = if lookup_unknown && exists {
let local_str = local_path.to_str().unwrap_or(".");
existing_promisor_state(local_str).await
} else {
PromisorProbe::NotPromisor
};
let mode = resolve_mirror_mode(withheld, exists, promisor);
// Surface the case where an unknown withheld-paths lookup kept (or, on an
// indeterminate probe, defensively applied) promisor mode instead of
// downgrading to a full clone. Derived from the resolved mode so it cannot
// drift from resolve_mirror_mode's preserve branch.
if lookup_unknown && mode == MirrorMode::Promisor {
warn!(
repo = %item.repo,
origin = %origin_url,
"withheld-paths lookup unavailable; using promisor mirror mode to avoid an unsafe full-clone downgrade"
);
}
let result = if local_path.exists() {
let result = if exists {
fetch_repo(&local_path, &remote_url, mode).await
} else {
clone_repo(&remote_url, &local_path, mode).await
@@ -197,6 +275,39 @@ async fn process_batch(
match result {
Ok(()) => {
info!(repo = %item.repo, origin = %origin_url, "synced repo from peer");
// iCaptcha propagation gate: on a first-time mirror, re-verify the
// proof the repo was created with (fetched from the origin) before
// admitting it. A node that enforces iCaptcha quarantines a mirror
// it cannot validate — kept on disk but hidden from serve/clone and
// listings until an operator releases it. Re-syncs of an already
// admitted repo keep their prior decision (upsert preserves it).
//
// "First-time" is keyed on DB-row absence, NOT on-disk presence: a
// prior attempt could have cloned to disk but crashed before the
// upsert, leaving no DB row. Disk-keying would skip admission on the
// retry and admit it unquarantined. On a DB lookup error, default to
// running the gate (fail toward quarantine, never silently admit).
let is_new_in_db = db
.get_repo(owner_short, repo_name)
.await
.map(|r| r.is_none())
.unwrap_or(true);
let quarantined = if is_new_in_db {
let proof =
fetch_icaptcha_proof(client, &origin_url, owner_short, repo_name).await;
match crate::icaptcha::admit_mirror(db, proof.as_deref(), owner_short).await {
crate::icaptcha::MirrorAdmission::Admit => false,
crate::icaptcha::MirrorAdmission::Quarantine(reason) => {
warn!(
repo = %item.repo, origin = %origin_url, reason,
"quarantining mirrored repo: failed iCaptcha propagation gate"
);
true
}
}
} else {
false
};
// Register in DB so git smart HTTP can serve the mirrored repo
let _ = db
.upsert_mirror_repo(
@@ -204,6 +315,7 @@ async fn process_batch(
repo_name,
local_path.to_str().unwrap_or(""),
machine_id,
quarantined,
)
.await;
// Option B2: carry the encrypted withheld-blob envelopes too, so an
@@ -247,7 +359,7 @@ async fn process_batch(
/// Query the origin's anonymous `withheld-paths` endpoint. Returns the withheld
/// glob list on a 2xx, or `None` on any non-success / network / parse error
/// (treated as "unknown" by `classify_mirror`).
/// (treated as "unknown" by `resolve_mirror_mode`).
async fn fetch_withheld(
client: &reqwest::Client,
origin_url: &str,
@@ -269,6 +381,25 @@ async fn fetch_withheld(
Some(globs)
}
/// Fetch the iCaptcha proof token the origin recorded for this repo, used by the
/// mirror-admission gate. Returns `None` on any non-success / network / parse
/// error, or when the origin has no proof for the repo (treated as "no proof" by
/// `icaptcha::admit_mirror`, which quarantines in enforce mode).
async fn fetch_icaptcha_proof(
client: &reqwest::Client,
origin_url: &str,
owner: &str,
repo: &str,
) -> Option<String> {
let url = format!("{origin_url}/api/v1/repos/{owner}/{repo}/icaptcha-proof");
let resp = client.get(&url).send().await.ok()?;
if !resp.status().is_success() {
return None;
}
let body: serde_json::Value = resp.json().await.ok()?;
body.get("proof")?.as_str().map(str::to_string)
}
/// Signed request path for replica registration on the origin node.
fn replica_registration_path(owner: &str, repo: &str) -> String {
format!("/api/v1/repos/{owner}/{repo}/replicas")
@@ -442,6 +573,38 @@ async fn git_config_get(repo: &str, key: &str) -> Option<String> {
(!value.is_empty()).then_some(value)
}
/// Probe an existing mirror's `remote.origin.promisor` config as a three-valued
/// state. Unlike [`git_config_get`], this distinguishes a definitively-absent key
/// from a probe failure so a transient git error cannot be read as "not a
/// promisor" and trigger a downgrade (issue #48).
///
/// `git config --get` exits 0 when the key is set, 1 when it is absent (or the
/// directory is not a git repo) — both definitive `NotPromisor` — and other
/// non-zero codes (e.g. 128 for a bad path or unreadable config) on error. A spawn
/// failure or any non-{0,1} exit is `Unknown`.
async fn existing_promisor_state(repo: &str) -> PromisorProbe {
let out = match tokio::process::Command::new("git")
.args(["-C", repo, "config", "--get", "remote.origin.promisor"])
.output()
.await
{
Ok(out) => out,
Err(_) => return PromisorProbe::Unknown,
};
match out.status.code() {
Some(0) => {
let value = String::from_utf8_lossy(&out.stdout).trim().to_string();
if value == "true" {
PromisorProbe::Promisor
} else {
PromisorProbe::NotPromisor
}
}
Some(1) => PromisorProbe::NotPromisor,
_ => PromisorProbe::Unknown,
}
}
/// Mirror-clone a repo from a remote URL into a local bare repo.
/// `Promisor` mode adds `--filter=blob:limit=10g`, which marks the repo a git
/// promisor (so a pack with origin-omitted withheld blobs is accepted) while
@@ -532,23 +695,66 @@ mod tests {
use tempfile::TempDir;
#[test]
fn classify_promisor_when_withheld_nonempty() {
let mode = classify_mirror(Some(vec!["/secret/**".to_string()]));
fn resolve_promisor_when_withheld_nonempty() {
let mode = resolve_mirror_mode(
Some(vec!["/secret/**".to_string()]),
true,
PromisorProbe::NotPromisor,
);
assert!(matches!(mode, MirrorMode::Promisor));
}
#[test]
fn classify_plain_when_withheld_empty() {
let mode = classify_mirror(Some(vec![]));
fn resolve_plain_when_withheld_empty() {
// A genuine public transition returns Some(empty) and still downgrades,
// regardless of whether the mirror exists or was a promisor.
for exists in [true, false] {
for probe in [
PromisorProbe::Promisor,
PromisorProbe::NotPromisor,
PromisorProbe::Unknown,
] {
let mode = resolve_mirror_mode(Some(vec![]), exists, probe);
assert!(matches!(mode, MirrorMode::Plain));
}
}
}
#[test]
fn resolve_preserves_promisor_on_unknown_lookup_for_existing_mirror() {
// Regression for #48: a transient withheld-paths outage (None) must NOT
// downgrade a still-withheld promisor mirror to a full clone.
let mode = resolve_mirror_mode(None, true, PromisorProbe::Promisor);
assert!(matches!(mode, MirrorMode::Promisor));
}
#[test]
fn resolve_preserves_promisor_on_indeterminate_probe() {
// Defense-in-depth (#48): if the config probe itself fails (Unknown) in the
// same cycle as a withheld-paths outage, bias toward preserving rather than
// firing the destructive downgrade.
let mode = resolve_mirror_mode(None, true, PromisorProbe::Unknown);
assert!(matches!(mode, MirrorMode::Promisor));
}
#[test]
fn resolve_plain_when_unknown_lookup_for_non_promisor_mirror() {
// An existing non-promisor mirror is unaffected by the preserve branch.
let mode = resolve_mirror_mode(None, true, PromisorProbe::NotPromisor);
assert!(matches!(mode, MirrorMode::Plain));
}
#[test]
fn classify_plain_when_lookup_failed() {
// None == 404 / network error / parse failure: attempt a plain mirror
// and let the git read endpoint fail-close a mode-A repo.
let mode = classify_mirror(None);
assert!(matches!(mode, MirrorMode::Plain));
fn resolve_plain_when_unknown_lookup_for_fresh_clone() {
// No local mirror yet: None stays Plain (fail-closed at the git layer).
for probe in [
PromisorProbe::Promisor,
PromisorProbe::NotPromisor,
PromisorProbe::Unknown,
] {
let mode = resolve_mirror_mode(None, false, probe);
assert!(matches!(mode, MirrorMode::Plain));
}
}
fn rb(oid: &str, cid: &str) -> ReplicaBlob {
@@ -690,6 +896,38 @@ mod tests {
assert_eq!(git_config(&dest, "remote.origin.mirror"), "true");
}
#[tokio::test]
async fn probe_reports_promisor_for_promisor_mirror() {
let (td, url) = bare_remote(&[("public/a.txt", b"pub\n")]);
let dest = td.path().join("mirror.git");
clone_repo(&url, &dest, MirrorMode::Promisor).await.unwrap();
let probe = existing_promisor_state(dest.to_str().unwrap()).await;
assert_eq!(probe, PromisorProbe::Promisor);
}
#[tokio::test]
async fn probe_reports_not_promisor_when_key_absent() {
// A plain mirror never sets remote.origin.promisor, so `git config --get`
// exits 1 (key absent) — the probe must read that as NotPromisor, never
// Unknown (which would wrongly preserve and upgrade a plain mirror).
let (td, url) = bare_remote(&[("public/a.txt", b"pub\n")]);
let dest = td.path().join("mirror.git");
clone_repo(&url, &dest, MirrorMode::Plain).await.unwrap();
let probe = existing_promisor_state(dest.to_str().unwrap()).await;
assert_eq!(probe, PromisorProbe::NotPromisor);
}
#[tokio::test]
async fn probe_reports_unknown_on_git_error() {
// A path git cannot resolve as a repo at all (exit 128) is an indeterminate
// probe, not a definitive "not a promisor" — it must map to Unknown so the
// caller preserves rather than downgrades (#48 defense-in-depth).
let probe = existing_promisor_state("/nonexistent/gitlawb-probe-xyz").await;
assert_eq!(probe, PromisorProbe::Unknown);
}
#[tokio::test]
async fn promisor_fetch_updates_existing_mirror() {
let (td, url) = bare_remote(&[("public/a.txt", b"pub\n")]);
@@ -755,4 +993,43 @@ mod tests {
register_replica_with_origin(&client, &keypair, Some(""), "http://127.0.0.1:1", "o", "r")
.await;
}
fn peer(did: &str, http_url: &str) -> crate::db::PeerRecord {
crate::db::PeerRecord {
did: did.to_string(),
http_url: http_url.to_string(),
last_seen: None,
last_ping_ok: false,
announced_at: String::new(),
}
}
#[test]
fn resolve_origin_url_matches_and_trims_trailing_slash() {
let peers = vec![
peer("did:key:a", "https://a.example/"),
peer("did:key:b", "https://b.example"),
];
// Trailing slash is trimmed so callers can append `/{path}` cleanly.
assert_eq!(
resolve_origin_url(&peers, "did:key:a").as_deref(),
Some("https://a.example")
);
// Already-trimmed URLs pass through unchanged.
assert_eq!(
resolve_origin_url(&peers, "did:key:b").as_deref(),
Some("https://b.example")
);
}
#[test]
fn resolve_origin_url_returns_none_for_unknown_did() {
let peers = vec![peer("did:key:a", "https://a.example")];
assert_eq!(resolve_origin_url(&peers, "did:key:unknown"), None);
}
#[test]
fn resolve_origin_url_returns_none_for_empty_peer_list() {
assert_eq!(resolve_origin_url(&[], "did:key:a"), None);
}
}
File diff suppressed because it is too large. Load diff
+67 -2
View File
@@ -5,6 +5,7 @@
//! `is_public` flag. It performs no I/O so it is exhaustively unit tested.
use crate::db::VisibilityRule;
use unicode_normalization::UnicodeNormalization;
#[derive(Debug, PartialEq, Eq)]
pub enum Decision {
@@ -12,6 +13,17 @@ pub enum Decision {
Deny,
}
/// NFC-normalize a glob prefix or path so the matcher compares canonically
/// equivalent strings byte-for-byte. Without this, a deny rule authored NFC
/// (`é` = U+00E9) byte-compares unequal to a path committed NFD (`e` + U+0301)
/// and the blob slips past the rule on the replication path (#101). NFC, not
/// NFKC: compatibility folding (ligatures, full-width forms) would merge paths
/// the filesystem treats as distinct and over-withhold. Both sides of every
/// comparison must pass through this, or the skew just moves.
fn nfc(s: &str) -> String {
s.nfc().collect()
}
/// True if `caller` is the repo owner (matches full did:key or its short form),
/// mirroring the owner-match idiom in `api/protect.rs`.
fn is_owner(owner_did: &str, caller: &str) -> bool {
@@ -36,12 +48,17 @@ fn glob_matches(glob: &str, path: &str) -> bool {
if prefix == "/" {
return true;
}
// Compare in NFC so an NFC rule matches a canonically-equivalent NFD path
// (and vice versa). Both sides normalized here — the single matcher seam.
let prefix = nfc(prefix);
let path = nfc(path);
path == prefix || path.starts_with(&format!("{prefix}/"))
}
/// Specificity = length of the match prefix; longer is more specific.
/// Specificity = length of the (normalized) match prefix; longer is more
/// specific. Normalized so ranking stays consistent with `glob_matches`.
fn specificity(glob: &str) -> usize {
glob_prefix(glob).len()
nfc(glob_prefix(glob)).len()
}
/// Decide whether `caller` (None = anonymous) may read `path` in a repo.
@@ -96,6 +113,20 @@ pub fn visibility_check(
}
}
/// Whether `caller` (None = anonymous) may see a repo in a listing — the `"/"`
/// visibility decision, shared by every repo-listing surface (REST list,
/// federated list, GraphQL `repos`) so they enforce one rule, not three drifting
/// copies. Not a bare `is_public` test: a repo can be `is_public=false` with a
/// root rule granting readers, or `is_public=true` with a root deny (#97).
pub fn listable_at_root(
rules: &[VisibilityRule],
is_public: bool,
owner_did: &str,
caller: Option<&str>,
) -> bool {
visibility_check(rules, is_public, owner_did, caller, "/") == Decision::Allow
}
/// The subtree path globs that `caller` (None = anonymous) may NOT read, given
/// the repo's rules. Whole-repo ("/") rules are excluded: a denied whole-repo
/// read is handled by the 404 gate before a clone ever starts. Each remaining
@@ -371,4 +402,38 @@ mod tests {
true
));
}
// #101: a deny rule must withhold a path that denotes the same characters in
// a different Unicode normalization form. Without NFC normalization at the
// matcher seam, an NFC-authored rule byte-compares unequal to an NFD-stored
// path and the blob leaks on the replication path.
#[test]
fn matcher_withholds_across_nfc_nfd_normalization_skew() {
// "é" composed (NFC, U+00E9) in the rule; decomposed (NFD, e + U+0301)
// in the committed path.
let nfc_rule = "/s\u{00e9}cret/**";
let nfd_path = "/se\u{0301}cret/key.pem";
let rules = [rule(nfc_rule, VisibilityMode::B, &["did:key:z6MkFriend"])];
assert_eq!(
visibility_check(&rules, true, OWNER, None, nfd_path),
Decision::Deny,
"NFC-authored deny rule must withhold the canonically-equivalent NFD path"
);
// Mirror: rule authored NFD, path committed NFC.
let nfd_rule = "/se\u{0301}cret/**";
let nfc_path = "/s\u{00e9}cret/key.pem";
let rules2 = [rule(nfd_rule, VisibilityMode::B, &["did:key:z6MkFriend"])];
assert_eq!(
visibility_check(&rules2, true, OWNER, None, nfc_path),
Decision::Deny,
"NFD-authored deny rule must withhold the canonically-equivalent NFC path"
);
// A genuinely different path is still allowed (no over-withholding).
assert_eq!(
visibility_check(&rules, true, OWNER, None, "/public/x.txt"),
Decision::Allow
);
}
}
+1 -1
View File
@@ -1,7 +1,7 @@
[package]
name = "gl"
description = "The gitlawb CLI — identity management, node control, MCP server"
version.workspace = true
version = "0.4.0" # x-release-please-version
edition.workspace = true
rust-version.workspace = true
license.workspace = true
+841 -67
View File
@@ -234,6 +234,22 @@ struct WithheldPathsResponse {
reinclude: Vec<String>,
}
/// The node's `/encrypted-blobs` 200 body. Same rationale as `WithheldPathsResponse`:
/// deserializing into a typed struct makes a missing or mistyped `blobs` field (or a
/// blob entry missing its `oid`) a hard error instead of silently becoming "nothing
/// to recover", which would mask a server schema regression behind a clone that
/// quietly omits authorized files. Unknown server fields (e.g. `size`) are ignored
/// by serde, so the response may carry extra keys without breaking.
#[derive(Deserialize)]
struct EncryptedBlobsResponse {
blobs: Vec<EncryptedBlobEntry>,
}
#[derive(Deserialize)]
struct EncryptedBlobEntry {
oid: String,
}
/// After the base clone, recover encrypted blobs the caller is authorized for
/// that are missing locally: fetch the envelope, decrypt with the caller's key,
/// install as a loose object. Returns the repo-relative paths recovered.
@@ -259,13 +275,8 @@ async fn recover_encrypted_blobs(
Ok(r) if r.status().is_success() => r,
_ => return Ok(vec![]),
};
let body: serde_json::Value = resp.json().await.context("parsing encrypted-blobs")?;
let blobs = body
.get("blobs")
.and_then(|b| b.as_array())
.cloned()
.unwrap_or_default();
if blobs.is_empty() {
let body: EncryptedBlobsResponse = resp.json().await.context("parsing encrypted-blobs")?;
if body.blobs.is_empty() {
return Ok(vec![]);
}
@@ -283,10 +294,8 @@ async fn recover_encrypted_blobs(
}
let mut recovered = Vec::new();
for entry in blobs {
let Some(oid) = entry.get("oid").and_then(|o| o.as_str()) else {
continue;
};
for entry in body.blobs {
let oid = entry.oid.as_str();
// Skip if already present locally.
let present = Command::new("git")
.args(["-C", dest_str, "cat-file", "-e", oid])
@@ -354,43 +363,151 @@ struct Manifest {
blobs: Vec<ManifestBlob>,
}
/// Extract transaction ids from an Arweave GraphQL `transactions` response.
fn parse_tx_ids(v: &serde_json::Value) -> Vec<String> {
v.get("data")
.and_then(|d| d.get("transactions"))
/// One discovered manifest transaction: its Arweave id and block height. The
/// height is `None` only when the anchor is not yet mined (no `block`), in which
/// case it is the most recent and sorts as newest in the merge tie-break. A mined
/// anchor is always `Some(_)` (0 if the gateway reported an unreadable height) so
/// a malformed height can never masquerade as pending/newest.
struct TxRef {
id: String,
height: Option<u64>,
}
/// Edges requested per Arweave GraphQL page (`first:` in the discovery query) and
/// the per-page bound enforced when parsing a response. The `first:N` literal in
/// `ARWEAVE_DISCOVERY_QUERY` MUST equal this constant; `query_first_matches_page_size`
/// pins that invariant at test time (the query keeps a literal because `format!`-ing
/// the brace-heavy GraphQL body would be error-prone).
const ARWEAVE_PAGE_SIZE: usize = 100;
/// Paginated discovery query: manifests tagged for this repo, newest first, with
/// `pageInfo` for cursor pagination and `block{height}` for the recency tie-break.
const ARWEAVE_DISCOVERY_QUERY: &str = r#"query($repo:String!,$cursor:String){transactions(tags:[{name:"App-Name",values:["gitlawb"]},{name:"Schema",values:["gitlawb/encrypted-manifest/v1"]},{name:"Repo",values:[$repo]}],first:100,after:$cursor){pageInfo{hasNextPage endCursor}edges{cursor node{id block{height}}}}}"#;
/// A parsed page of an Arweave GraphQL `transactions` response: the edge refs
/// plus pagination state. `has_next` is `None` when the response omitted
/// `pageInfo` entirely (an older or partial gateway), which the discovery loop
/// treats as terminal but flags when the page was full (possible truncation).
struct TxPage {
refs: Vec<TxRef>,
has_next: Option<bool>,
end_cursor: Option<String>,
}
/// Parse one Arweave GraphQL `transactions` page: each edge's tx id and block
/// height, plus `pageInfo` for cursor pagination. Hand-walks the `Value` (the
/// envelope nests `data.transactions.{edges,pageInfo}`) rather than deriving a
/// wrapper struct, matching the surrounding style.
///
/// At most `ARWEAVE_PAGE_SIZE` edges are taken from a single page: the query asks
/// for `first:100`, so a response with more edges is a misbehaving (or hostile)
/// gateway. Bounding here caps the per-page `TxRef` vec and the downstream per-id
/// fetch loop so the `MAX_TX_IDS` budget can't be defeated by one oversized page.
/// (It does not bound the response body itself: `reqwest` buffers and deserializes
/// the whole body before this runs — an unbounded-body concern shared by every
/// gateway read in this file, addressed separately if it ever matters.)
fn parse_tx_page(v: &serde_json::Value) -> TxPage {
let txns = v.get("data").and_then(|d| d.get("transactions"));
let refs: Vec<TxRef> = txns
.and_then(|t| t.get("edges"))
.and_then(|e| e.as_array())
.map(|edges| {
edges
.iter()
.take(ARWEAVE_PAGE_SIZE)
.filter_map(|edge| {
edge.get("node")
.and_then(|n| n.get("id"))
.and_then(|i| i.as_str())
.map(String::from)
let node = edge.get("node")?;
let id = node.get("id").and_then(|i| i.as_str())?.to_string();
// Distinguish "not yet mined" from "mined but height unreadable".
// Only a genuinely absent/null `block` is pending (`None`,
// ranked newest). A present `block` means the anchor IS mined, so
// it must not be promoted to newest: parse the height permissively
// (number or numeric string, guarding a gateway that stringifies
// large ints), and an unreadable height ranks lowest (0) rather
// than collapsing to `None`/newest.
let height = match node.get("block") {
None | Some(serde_json::Value::Null) => None,
Some(b) => Some(
b.get("height")
.and_then(|h| {
h.as_u64()
.or_else(|| h.as_str().and_then(|s| s.parse::<u64>().ok()))
})
.unwrap_or(0),
),
};
Some(TxRef { id, height })
})
.collect()
})
.unwrap_or_default()
.unwrap_or_default();
let page_info = txns.and_then(|t| t.get("pageInfo"));
let has_next = page_info
.and_then(|p| p.get("hasNextPage"))
.and_then(|h| h.as_bool());
let end_cursor = page_info
.and_then(|p| p.get("endCursor"))
.and_then(|c| c.as_str())
.map(String::from);
TxPage {
refs,
has_next,
end_cursor,
}
}
/// Merge per-push manifests into a single `oid -> cid` map, latest-wins by the
/// manifest `timestamp` (RFC3339, compared lexicographically; a later push that
/// re-sealed a blob overrides the earlier entry).
fn merge_manifests(manifests: Vec<Manifest>) -> std::collections::HashMap<String, String> {
let mut best: std::collections::HashMap<String, (String, String)> =
std::collections::HashMap::new(); // oid -> (cid, timestamp)
for m in manifests {
/// Order a block height so that a not-yet-mined anchor (`None`) sorts above any
/// mined one, and mined anchors order by ascending height. Higher rank == newer.
fn height_rank(h: Option<u64>) -> (u8, u64) {
match h {
None => (1, 0), // pending anchor: newest
Some(v) => (0, v), // mined: by block height
}
}
/// Merge per-push manifests into a single `oid -> cid` map. A later seal of an
/// oid overrides an earlier one, ranked by the composite key
/// `(timestamp, block height, cid)`:
/// - manifest `timestamp` (RFC3339, lexicographic) is primary — the unchanged
/// latest-wins behavior;
/// - Arweave block height breaks equal-timestamp ties by anchor recency (a
/// pending anchor counts as newest), preserving the "newest wins" that
/// single-page `HEIGHT_DESC` discovery used to provide by insertion order.
/// The height is gateway-reported (chain-derived but not independently
/// verified here), so it removes order-dependence and honest-node clock skew,
/// not trust in the gateway — discovery already assumes an honest gateway;
/// - cid is a final lexicographic tiebreak so two anchors in the *same* block
/// still resolve deterministically (arbitrary, not recency-ordered, within a
/// block), independent of discovery order.
fn merge_manifests(
manifests: Vec<(Manifest, Option<u64>)>,
) -> std::collections::HashMap<String, String> {
// oid -> (cid, timestamp, height_rank)
let mut best: std::collections::HashMap<String, (String, String, (u8, u64))> =
std::collections::HashMap::new();
for (m, height) in manifests {
let cand_rank = height_rank(height);
for b in m.blobs {
match best.get(&b.oid) {
Some((_, ts)) if ts.as_str() >= m.timestamp.as_str() => {}
_ => {
best.insert(b.oid, (b.cid, m.timestamp.clone()));
let keep_incumbent = match best.get(&b.oid) {
// Incumbent wins iff its composite key is >= the candidate's. The
// `>=` (not `>`) keeps the incumbent only on a full tie, where the
// entries are identical anyway. A strictly-greater candidate key
// (newer timestamp, or equal timestamp + higher block, or equal
// both + larger cid) replaces it.
Some((cur_cid, cur_ts, cur_rank)) => {
(cur_ts.as_str(), cur_rank, cur_cid.as_str())
>= (m.timestamp.as_str(), &cand_rank, b.cid.as_str())
}
None => false,
};
if !keep_incumbent {
best.insert(b.oid, (b.cid, m.timestamp.clone(), cand_rank));
}
}
}
best.into_iter().map(|(oid, (cid, _))| (oid, cid)).collect()
best.into_iter()
.map(|(oid, (cid, _, _))| (oid, cid))
.collect()
}
/// Option B3 fallback recovery, with no dependency on a gitlawb node API. Query
@@ -424,36 +541,111 @@ async fn recover_from_arweave(
.build()
.unwrap_or_else(|_| reqwest::Client::new());
// 1. Discover manifest transaction ids via Arweave GraphQL.
let query = r#"query($repo:String!){transactions(tags:[{name:"App-Name",values:["gitlawb"]},{name:"Schema",values:["gitlawb/encrypted-manifest/v1"]},{name:"Repo",values:[$repo]}],first:100){edges{node{id}}}}"#;
let gql_body = serde_json::json!({ "query": query, "variables": { "repo": slug } });
let resp = match client
.post(format!("{ag}/graphql"))
.json(&gql_body)
.send()
.await
{
Ok(r) if r.status().is_success() => r,
_ => return Ok(vec![]),
};
let gql: serde_json::Value = match resp.json().await {
Ok(v) => v,
Err(_) => return Ok(vec![]),
};
let tx_ids = parse_tx_ids(&gql);
if tx_ids.is_empty() {
// 1. Discover manifest transactions via paginated Arweave GraphQL. Collect
// (tx id, block height) refs across every page, deduping by id and bounding
// the loop so a hostile or buggy gateway cannot grow the set, the page
// count, or the downstream per-id fetch loop without limit. Best-effort: a
// page-1 failure (gateway down/unconfigured) stays silent as before; a
// failure or cap reached *after* discovery began warns, since the recovery
// is then knowingly partial.
const MAX_PAGES: usize = 1000;
const MAX_TX_IDS: usize = 10_000;
let query = ARWEAVE_DISCOVERY_QUERY;
let mut after: Option<String> = None;
let mut seen: std::collections::HashSet<String> = std::collections::HashSet::new();
let mut refs: Vec<TxRef> = Vec::new();
let mut prev_cursor: Option<String> = None;
let mut page_num = 0usize;
loop {
if page_num >= MAX_PAGES {
eprintln!(
"warning: Arweave manifest discovery hit the {MAX_PAGES}-page cap; \
some authorized files may not be recovered"
);
break;
}
page_num += 1;
let gql_body =
serde_json::json!({ "query": query, "variables": { "repo": slug, "cursor": after } });
// Any gateway failure (non-2xx, send error, unparseable body) ends the loop.
let response = match client
.post(format!("{ag}/graphql"))
.json(&gql_body)
.send()
.await
{
Ok(r) if r.status().is_success() => r.json::<serde_json::Value>().await.ok(),
_ => None,
};
let value = match response {
Some(v) => v,
None => {
// A failure *after* discovery began means recovery is knowingly
// partial, so warn. A page-1 failure is the benign unconfigured/
// unreachable-gateway case and stays silent, exactly as before.
if !refs.is_empty() {
eprintln!(
"warning: Arweave manifest discovery was interrupted by a gateway \
error; some authorized files may not be recovered"
);
}
break;
}
};
let page = parse_tx_page(&value);
let full = page.refs.len() >= ARWEAVE_PAGE_SIZE;
for r in page.refs {
if seen.insert(r.id.clone()) {
refs.push(r);
}
}
// Cap checked at the page boundary so a partial page is never half-merged.
if refs.len() >= MAX_TX_IDS {
eprintln!(
"warning: Arweave manifest discovery hit the {MAX_TX_IDS}-transaction cap; \
some authorized files may not be recovered"
);
break;
}
match page.has_next {
Some(true) => match page.end_cursor {
// Advance only on a fresh cursor; a missing or repeating cursor is
// a degenerate gateway response and must not loop forever.
Some(c) if Some(&c) != prev_cursor.as_ref() => {
prev_cursor = Some(c.clone());
after = Some(c);
}
_ => break,
},
Some(false) => break,
// No pageInfo: terminal. A *full* page without it may be a silently
// truncated gateway, so flag that; a short page is genuinely the end.
None => {
if full {
eprintln!(
"warning: Arweave gateway returned a full page without pagination \
metadata; discovery may be incomplete and some authorized files may \
not be recovered"
);
}
break;
}
}
}
if refs.is_empty() {
return Ok(vec![]);
}
// 2. Fetch and parse each manifest body, then merge latest-wins per oid.
let mut manifests = Vec::new();
for tx in tx_ids {
let m = match client.get(format!("{ag}/{tx}")).send().await {
Ok(r) if r.status().is_success() => r,
// 2. Fetch each manifest body, pair it with its anchor's block height, then
// merge (latest-wins by timestamp, tie-broken by block height then cid).
let mut manifests: Vec<(Manifest, Option<u64>)> = Vec::new();
for r in refs {
let m = match client.get(format!("{ag}/{}", r.id)).send().await {
Ok(resp) if resp.status().is_success() => resp,
_ => continue,
};
if let Ok(parsed) = m.json::<Manifest>().await {
manifests.push(parsed);
manifests.push((parsed, r.height));
}
}
let oid_cid = merge_manifests(manifests);
@@ -547,7 +739,14 @@ pub async fn run(args: CloneArgs) -> Result<()> {
// fallback for any authorized blobs the node could not supply.
let mut paths = recover_encrypted_blobs(&args.node, &owner, &name, &dest, &keypair)
.await
.unwrap_or_default();
.unwrap_or_else(|e| {
// A node-recovery error must not abort the clone (the Arweave
// fallback still runs), but the strict /encrypted-blobs parse now
// fails closed on schema drift, so surface it rather than letting
// `.unwrap_or_default()` silently swallow it into "no paths".
eprintln!("warning: encrypted-blobs recovery failed: {e}");
Vec::new()
});
let from_arweave = recover_from_arweave(
&args.arweave_gateway,
&args.ipfs_gateway,
@@ -788,19 +987,113 @@ mod tests {
}
#[test]
fn parse_tx_ids_extracts_node_ids() {
let v: serde_json::Value = serde_json::from_str(
r#"{"data":{"transactions":{"edges":[{"node":{"id":"TX1"}},{"node":{"id":"TX2"}}]}}}"#,
)
.unwrap();
assert_eq!(parse_tx_ids(&v), vec!["TX1".to_string(), "TX2".to_string()]);
fn encrypted_blobs_response_is_schema_strict() {
// Happy path: a well-formed body parses and exposes the oids.
let ok: EncryptedBlobsResponse =
serde_json::from_str(r#"{"blobs":[{"oid":"abc"}]}"#).unwrap();
assert_eq!(ok.blobs.len(), 1);
assert_eq!(ok.blobs[0].oid, "abc");
// Unknown server fields on an entry are tolerated (no deny_unknown_fields).
let extra: EncryptedBlobsResponse =
serde_json::from_str(r#"{"blobs":[{"oid":"abc","size":42}]}"#).unwrap();
assert_eq!(extra.blobs[0].oid, "abc");
// Empty list is valid and distinct from a missing field.
let empty: EncryptedBlobsResponse = serde_json::from_str(r#"{"blobs":[]}"#).unwrap();
assert!(empty.blobs.is_empty());
// Schema drift is a hard error, not a silent "nothing to recover":
// missing `blobs`, wrong-typed `blobs`, and an entry missing `oid`.
assert!(serde_json::from_str::<EncryptedBlobsResponse>(r#"{"items":[]}"#).is_err());
assert!(serde_json::from_str::<EncryptedBlobsResponse>(r#"{"blobs":"nope"}"#).is_err());
assert!(
serde_json::from_str::<EncryptedBlobsResponse>(r#"{"blobs":[{"size":1}]}"#).is_err()
);
}
#[test]
fn parse_tx_ids_empty_on_no_edges() {
fn parse_tx_page_extracts_node_ids_and_heights() {
let v: serde_json::Value = serde_json::from_str(
r#"{"data":{"transactions":{"pageInfo":{"hasNextPage":true,"endCursor":"C1"},"edges":[{"node":{"id":"TX1","block":{"height":100}}},{"node":{"id":"TX2","block":null}}]}}}"#,
)
.unwrap();
let page = parse_tx_page(&v);
let ids: Vec<&str> = page.refs.iter().map(|r| r.id.as_str()).collect();
assert_eq!(ids, vec!["TX1", "TX2"]);
assert_eq!(page.refs[0].height, Some(100));
assert_eq!(page.refs[1].height, None); // null block -> pending -> None
assert_eq!(page.has_next, Some(true));
assert_eq!(page.end_cursor.as_deref(), Some("C1"));
}
#[test]
fn parse_tx_page_empty_on_no_edges() {
let v: serde_json::Value =
serde_json::from_str(r#"{"data":{"transactions":{"edges":[]}}}"#).unwrap();
assert!(parse_tx_ids(&v).is_empty());
let page = parse_tx_page(&v);
assert!(page.refs.is_empty());
// No pageInfo present -> has_next None (the loop treats it as terminal).
assert_eq!(page.has_next, None);
assert_eq!(page.end_cursor, None);
}
#[test]
fn parse_tx_page_missing_block_is_none() {
let v: serde_json::Value =
serde_json::from_str(r#"{"data":{"transactions":{"edges":[{"node":{"id":"TX1"}}]}}}"#)
.unwrap();
let page = parse_tx_page(&v);
assert_eq!(page.refs[0].height, None);
}
#[test]
fn parse_tx_page_height_parsing_is_robust() {
// A gateway that serializes height as a JSON string (a habit to dodge
// 53-bit float precision) must still parse to the real height, not
// collapse to None and masquerade as a pending/newest anchor.
let v: serde_json::Value = serde_json::from_str(
r#"{"data":{"transactions":{"edges":[
{"node":{"id":"TXSTR","block":{"height":"12345"}}},
{"node":{"id":"TXBAD","block":{"height":"not-a-number"}}},
{"node":{"id":"TXNOH","block":{}}}
]}}}"#,
)
.unwrap();
let page = parse_tx_page(&v);
// Numeric string parses to the real height.
assert_eq!(page.refs[0].height, Some(12345));
// A present block with an unreadable/absent height is mined-but-unknown:
// it ranks lowest (Some(0)), NOT None/newest, so it can't steal a tie.
assert_eq!(page.refs[1].height, Some(0));
assert_eq!(page.refs[2].height, Some(0));
}
#[test]
fn parse_tx_page_bounds_edges_per_page() {
// A hostile gateway returning more than ARWEAVE_PAGE_SIZE edges on one
// page must not defeat the MAX_TX_IDS budget: parse caps per-page refs so
// the downstream fetch loop stays bounded.
let edges: String = (0..ARWEAVE_PAGE_SIZE + 50)
.map(|i| format!(r#"{{"node":{{"id":"TX{i}"}}}}"#))
.collect::<Vec<_>>()
.join(",");
let v: serde_json::Value = serde_json::from_str(&format!(
r#"{{"data":{{"transactions":{{"edges":[{edges}]}}}}}}"#
))
.unwrap();
let page = parse_tx_page(&v);
assert_eq!(page.refs.len(), ARWEAVE_PAGE_SIZE);
}
#[test]
fn query_first_matches_page_size() {
// The `first:N` the discovery query requests must equal the per-page bound
// enforced in parse_tx_page, or full-page detection and the cap drift.
assert!(
ARWEAVE_DISCOVERY_QUERY.contains(&format!("first:{ARWEAVE_PAGE_SIZE}")),
"discovery query `first:` must equal ARWEAVE_PAGE_SIZE ({ARWEAVE_PAGE_SIZE})"
);
}
#[test]
@@ -837,7 +1130,7 @@ mod tests {
},
],
};
let merged = merge_manifests(vec![older, newer]);
let merged = merge_manifests(vec![(older, Some(1)), (newer, Some(2))]);
assert_eq!(merged.get("o1").map(String::as_str), Some("cidNEW"));
assert_eq!(merged.get("o2").map(String::as_str), Some("cid2"));
}
@@ -859,10 +1152,109 @@ mod tests {
}],
};
// Newer first, older second: newer must still win.
let merged = merge_manifests(vec![newer, older]);
let merged = merge_manifests(vec![(newer, Some(2)), (older, Some(1))]);
assert_eq!(merged.get("o1").map(String::as_str), Some("cidNEW"));
}
// Helper: a single-blob manifest for the merge tie-break tests.
fn manifest_with(ts: &str, oid: &str, cid: &str) -> Manifest {
Manifest {
timestamp: ts.to_string(),
blobs: vec![ManifestBlob {
oid: oid.to_string(),
cid: cid.to_string(),
}],
}
}
#[test]
fn merge_equal_timestamp_breaks_by_block_height() {
// Same timestamp, different block heights: the higher block (newer
// anchor) must win, regardless of input order.
let lo = manifest_with("2026-06-11T00:00:00Z", "o1", "cidLO");
let hi = manifest_with("2026-06-11T00:00:00Z", "o1", "cidHI");
let a = merge_manifests(vec![
(
manifest_with("2026-06-11T00:00:00Z", "o1", "cidLO"),
Some(10),
),
(
manifest_with("2026-06-11T00:00:00Z", "o1", "cidHI"),
Some(20),
),
]);
assert_eq!(a.get("o1").map(String::as_str), Some("cidHI"));
// Reversed order, same result.
let b = merge_manifests(vec![(hi, Some(20)), (lo, Some(10))]);
assert_eq!(b.get("o1").map(String::as_str), Some("cidHI"));
}
#[test]
fn merge_pending_anchor_is_newest_on_tie() {
// Same timestamp; one mined anchor, one pending (None height). The
// pending one is the most recent and must win, order-independently.
let mined = manifest_with("2026-06-11T00:00:00Z", "o1", "cidMINED");
let pending = manifest_with("2026-06-11T00:00:00Z", "o1", "cidPENDING");
let a = merge_manifests(vec![(mined, Some(99)), (pending, None)]);
assert_eq!(a.get("o1").map(String::as_str), Some("cidPENDING"));
let mined2 = manifest_with("2026-06-11T00:00:00Z", "o1", "cidMINED");
let pending2 = manifest_with("2026-06-11T00:00:00Z", "o1", "cidPENDING");
let b = merge_manifests(vec![(pending2, None), (mined2, Some(99))]);
assert_eq!(b.get("o1").map(String::as_str), Some("cidPENDING"));
}
#[test]
fn merge_same_block_breaks_by_cid_deterministically() {
// Equal timestamp AND equal height: the cid tiebreak makes the result
// deterministic and order-independent (lexicographically-larger cid wins).
let a = merge_manifests(vec![
(
manifest_with("2026-06-11T00:00:00Z", "o1", "cidAAA"),
Some(5),
),
(
manifest_with("2026-06-11T00:00:00Z", "o1", "cidBBB"),
Some(5),
),
]);
let b = merge_manifests(vec![
(
manifest_with("2026-06-11T00:00:00Z", "o1", "cidBBB"),
Some(5),
),
(
manifest_with("2026-06-11T00:00:00Z", "o1", "cidAAA"),
Some(5),
),
]);
assert_eq!(a.get("o1"), b.get("o1"));
assert_eq!(a.get("o1").map(String::as_str), Some("cidBBB"));
}
#[test]
fn merge_newer_timestamp_wins_regardless_of_height() {
// Timestamp is primary: a newer timestamp wins even with a lower block
// height than the older-timestamp manifest.
let older_hi_block = manifest_with("2026-06-10T00:00:00Z", "o1", "cidOLD");
let newer_lo_block = manifest_with("2026-06-11T00:00:00Z", "o1", "cidNEW");
let merged = merge_manifests(vec![
(older_hi_block, Some(9999)),
(newer_lo_block, Some(1)),
]);
assert_eq!(merged.get("o1").map(String::as_str), Some("cidNEW"));
}
#[test]
fn merge_empty_timestamp_loses_to_timestamped() {
// A manifest with a missing/empty timestamp (serde default) sorts lowest
// and loses to any timestamped manifest for the same oid. Pins the
// empty-sorts-lowest precondition as a deliberate decision.
let untimestamped = manifest_with("", "o1", "cidEMPTY");
let timestamped = manifest_with("2026-06-11T00:00:00Z", "o1", "cidTS");
let a = merge_manifests(vec![(untimestamped, Some(50)), (timestamped, Some(1))]);
assert_eq!(a.get("o1").map(String::as_str), Some("cidTS"));
}
/// Read-path end-to-end over a mocked Arweave + IPFS gateway: discover the
/// manifest via GraphQL, fetch it, fetch the envelope, decrypt with the
/// caller's key, and install the previously-withheld blob.
@@ -961,6 +1353,388 @@ mod tests {
);
}
/// Discovery must follow cursor pagination: a blob whose manifest is anchored
/// only on the SECOND page is still recovered. If the loop stopped at page 1
/// (the old `first:100` behavior), this blob would be lost.
#[tokio::test]
async fn recover_from_arweave_paginates_to_later_page() {
use gitlawb_core::encrypt::seal_blob;
use gitlawb_core::identity::Keypair;
let (td, url) = bare_remote(&[("public/a.txt", b"pub\n"), ("secret/b.txt", b"SECRET\n")]);
let dest = td.path().join("dest");
let bare = url.strip_prefix("file://").unwrap();
assert!(Command::new("git")
.args(["-C", bare, "config", "uploadpack.allowFilter", "true"])
.status()
.unwrap()
.success());
setup_partial_clone(&dest, &url, &["/secret/**".to_string()], &[], None).unwrap();
let oid = {
let out = Command::new("git")
.args([
"-C",
dest.to_str().unwrap(),
"rev-parse",
"HEAD:secret/b.txt",
])
.output()
.unwrap();
String::from_utf8_lossy(&out.stdout).trim().to_string()
};
std::fs::remove_dir_all(url.strip_prefix("file://").unwrap()).unwrap();
let reader = Keypair::generate();
let envelope = seal_blob(b"SECRET\n", &[reader.verifying_key()]).unwrap();
let cid = "testcid123";
let mut server = mockito::Server::new_async().await;
// Page 1 (cursor null): hasNextPage=true, endCursor C1, an empty manifest
// anchor that does NOT carry the withheld blob.
let _gql_p1 = server
.mock("POST", "/graphql")
.match_body(mockito::Matcher::PartialJsonString(
r#"{"variables":{"cursor":null}}"#.into(),
))
.with_status(200)
.with_header("content-type", "application/json")
.with_body(
r#"{"data":{"transactions":{"pageInfo":{"hasNextPage":true,"endCursor":"C1"},"edges":[{"node":{"id":"TXP1","block":{"height":10}}}]}}}"#,
)
.create_async()
.await;
// Page 2 (cursor "C1"): hasNextPage=false, the anchor with the blob.
let _gql_p2 = server
.mock("POST", "/graphql")
.match_body(mockito::Matcher::PartialJsonString(
r#"{"variables":{"cursor":"C1"}}"#.into(),
))
.with_status(200)
.with_header("content-type", "application/json")
.with_body(
r#"{"data":{"transactions":{"pageInfo":{"hasNextPage":false,"endCursor":"C2"},"edges":[{"node":{"id":"TX1","block":{"height":20}}}]}}}"#,
)
.create_async()
.await;
let _tx_p1 = server
.mock("GET", "/TXP1")
.with_status(200)
.with_header("content-type", "application/json")
.with_body(r#"{"timestamp":"2026-06-10T00:00:00Z","blobs":[]}"#)
.create_async()
.await;
let manifest_body = serde_json::json!({
"timestamp": "2026-06-11T00:00:00Z",
"blobs": [{ "oid": oid, "cid": cid, "recipients": [] }],
})
.to_string();
let _tx1 = server
.mock("GET", "/TX1")
.with_status(200)
.with_header("content-type", "application/json")
.with_body(manifest_body)
.create_async()
.await;
let _blob = server
.mock("GET", format!("/ipfs/{cid}").as_str())
.with_status(200)
.with_body(envelope)
.create_async()
.await;
let paths = recover_from_arweave(
&server.url(),
&server.url(),
"alice",
"myrepo",
&dest,
&reader,
)
.await
.unwrap();
assert_eq!(
paths,
vec!["secret/b.txt".to_string()],
"blob anchored only on page 2 must be recovered via pagination"
);
}
/// Best-effort on mid-pagination failure: page 1 already yielded a usable
/// anchor; page 2 fails. Recovery still installs page 1's blob (partial Ok),
/// never aborts the clone.
#[tokio::test]
async fn recover_from_arweave_partial_on_midpagination_failure() {
use gitlawb_core::encrypt::seal_blob;
use gitlawb_core::identity::Keypair;
let (td, url) = bare_remote(&[("public/a.txt", b"pub\n"), ("secret/b.txt", b"SECRET\n")]);
let dest = td.path().join("dest");
let bare = url.strip_prefix("file://").unwrap();
assert!(Command::new("git")
.args(["-C", bare, "config", "uploadpack.allowFilter", "true"])
.status()
.unwrap()
.success());
setup_partial_clone(&dest, &url, &["/secret/**".to_string()], &[], None).unwrap();
let oid = {
let out = Command::new("git")
.args([
"-C",
dest.to_str().unwrap(),
"rev-parse",
"HEAD:secret/b.txt",
])
.output()
.unwrap();
String::from_utf8_lossy(&out.stdout).trim().to_string()
};
std::fs::remove_dir_all(url.strip_prefix("file://").unwrap()).unwrap();
let reader = Keypair::generate();
let envelope = seal_blob(b"SECRET\n", &[reader.verifying_key()]).unwrap();
let cid = "testcid123";
let mut server = mockito::Server::new_async().await;
// Page 1 (cursor null): the blob anchor, hasNextPage=true.
let manifest_body = serde_json::json!({
"timestamp": "2026-06-11T00:00:00Z",
"blobs": [{ "oid": oid, "cid": cid, "recipients": [] }],
})
.to_string();
let _gql_p1 = server
.mock("POST", "/graphql")
.match_body(mockito::Matcher::PartialJsonString(
r#"{"variables":{"cursor":null}}"#.into(),
))
.with_status(200)
.with_header("content-type", "application/json")
.with_body(
r#"{"data":{"transactions":{"pageInfo":{"hasNextPage":true,"endCursor":"C1"},"edges":[{"node":{"id":"TX1","block":{"height":20}}}]}}}"#,
)
.create_async()
.await;
// Page 2 (cursor "C1"): gateway error.
let _gql_p2 = server
.mock("POST", "/graphql")
.match_body(mockito::Matcher::PartialJsonString(
r#"{"variables":{"cursor":"C1"}}"#.into(),
))
.with_status(500)
.create_async()
.await;
let _tx1 = server
.mock("GET", "/TX1")
.with_status(200)
.with_header("content-type", "application/json")
.with_body(manifest_body)
.create_async()
.await;
let _blob = server
.mock("GET", format!("/ipfs/{cid}").as_str())
.with_status(200)
.with_body(envelope)
.create_async()
.await;
let paths = recover_from_arweave(
&server.url(),
&server.url(),
"alice",
"myrepo",
&dest,
&reader,
)
.await
.expect("mid-pagination failure must not abort recovery");
assert_eq!(
paths,
vec!["secret/b.txt".to_string()],
"page-1 blob must still be recovered when page 2 fails"
);
}
/// A gateway that reports `hasNextPage=true` but never advances the cursor
/// (returns the same `endCursor` every page) must not loop forever: the
/// degenerate-cursor guard terminates the loop and recovery returns based on
/// what was seen. The test completing at all proves no hang.
#[tokio::test]
async fn recover_from_arweave_terminates_on_nonadvancing_cursor() {
use gitlawb_core::encrypt::seal_blob;
use gitlawb_core::identity::Keypair;
let (td, url) = bare_remote(&[("public/a.txt", b"pub\n"), ("secret/b.txt", b"SECRET\n")]);
let dest = td.path().join("dest");
let bare = url.strip_prefix("file://").unwrap();
assert!(Command::new("git")
.args(["-C", bare, "config", "uploadpack.allowFilter", "true"])
.status()
.unwrap()
.success());
setup_partial_clone(&dest, &url, &["/secret/**".to_string()], &[], None).unwrap();
let oid = {
let out = Command::new("git")
.args([
"-C",
dest.to_str().unwrap(),
"rev-parse",
"HEAD:secret/b.txt",
])
.output()
.unwrap();
String::from_utf8_lossy(&out.stdout).trim().to_string()
};
std::fs::remove_dir_all(url.strip_prefix("file://").unwrap()).unwrap();
let reader = Keypair::generate();
let envelope = seal_blob(b"SECRET\n", &[reader.verifying_key()]).unwrap();
let cid = "testcid123";
let mut server = mockito::Server::new_async().await;
// Every GraphQL POST returns hasNextPage=true with the SAME endCursor.
// The guard fires on the second page (cursor repeats prev_cursor) and
// breaks, so discovery makes exactly two POSTs: page 1 (cursor null) records
// STUCK, page 2 (cursor STUCK) sees the repeat and stops. We assert that
// exact count below — a regression in the advance logic that loops extra
// times in the happy path fails the expect(2). (MAX_PAGES is the hard
// infinite-loop backstop independent of this guard.)
let manifest_body = serde_json::json!({
"timestamp": "2026-06-11T00:00:00Z",
"blobs": [{ "oid": oid, "cid": cid, "recipients": [] }],
})
.to_string();
let gql = server
.mock("POST", "/graphql")
.expect(2)
.with_status(200)
.with_header("content-type", "application/json")
.with_body(
r#"{"data":{"transactions":{"pageInfo":{"hasNextPage":true,"endCursor":"STUCK"},"edges":[{"node":{"id":"TX1","block":{"height":20}}}]}}}"#,
)
.create_async()
.await;
let _tx1 = server
.mock("GET", "/TX1")
.with_status(200)
.with_header("content-type", "application/json")
.with_body(manifest_body)
.create_async()
.await;
let _blob = server
.mock("GET", format!("/ipfs/{cid}").as_str())
.with_status(200)
.with_body(envelope)
.create_async()
.await;
let paths = recover_from_arweave(
&server.url(),
&server.url(),
"alice",
"myrepo",
&dest,
&reader,
)
.await
.expect("non-advancing cursor must terminate, not hang or error");
assert_eq!(paths, vec!["secret/b.txt".to_string()]);
gql.assert_async().await; // exactly two discovery POSTs, then the guard breaks
}
/// A tx id repeated across pages must be fetched only once: the cross-page
/// dedup (R8) bounds the downstream per-id manifest fetch loop.
#[tokio::test]
async fn recover_from_arweave_dedups_tx_ids_across_pages() {
use gitlawb_core::encrypt::seal_blob;
use gitlawb_core::identity::Keypair;
let (td, url) = bare_remote(&[("public/a.txt", b"pub\n"), ("secret/b.txt", b"SECRET\n")]);
let dest = td.path().join("dest");
let bare = url.strip_prefix("file://").unwrap();
assert!(Command::new("git")
.args(["-C", bare, "config", "uploadpack.allowFilter", "true"])
.status()
.unwrap()
.success());
setup_partial_clone(&dest, &url, &["/secret/**".to_string()], &[], None).unwrap();
let oid = {
let out = Command::new("git")
.args([
"-C",
dest.to_str().unwrap(),
"rev-parse",
"HEAD:secret/b.txt",
])
.output()
.unwrap();
String::from_utf8_lossy(&out.stdout).trim().to_string()
};
std::fs::remove_dir_all(url.strip_prefix("file://").unwrap()).unwrap();
let reader = Keypair::generate();
let envelope = seal_blob(b"SECRET\n", &[reader.verifying_key()]).unwrap();
let cid = "testcid123";
let mut server = mockito::Server::new_async().await;
// Page 1 and page 2 both list TX1 (cursor advances); the dedup must fetch
// the /TX1 manifest exactly once.
let _gql_p1 = server
.mock("POST", "/graphql")
.match_body(mockito::Matcher::PartialJsonString(
r#"{"variables":{"cursor":null}}"#.into(),
))
.with_status(200)
.with_header("content-type", "application/json")
.with_body(
r#"{"data":{"transactions":{"pageInfo":{"hasNextPage":true,"endCursor":"C1"},"edges":[{"node":{"id":"TX1","block":{"height":20}}}]}}}"#,
)
.create_async()
.await;
let _gql_p2 = server
.mock("POST", "/graphql")
.match_body(mockito::Matcher::PartialJsonString(
r#"{"variables":{"cursor":"C1"}}"#.into(),
))
.with_status(200)
.with_header("content-type", "application/json")
.with_body(
r#"{"data":{"transactions":{"pageInfo":{"hasNextPage":false,"endCursor":"C2"},"edges":[{"node":{"id":"TX1","block":{"height":20}}}]}}}"#,
)
.create_async()
.await;
let manifest_body = serde_json::json!({
"timestamp": "2026-06-11T00:00:00Z",
"blobs": [{ "oid": oid, "cid": cid, "recipients": [] }],
})
.to_string();
let tx1 = server
.mock("GET", "/TX1")
.expect(1) // dedup: fetched once despite appearing on both pages
.with_status(200)
.with_header("content-type", "application/json")
.with_body(manifest_body)
.create_async()
.await;
let _blob = server
.mock("GET", format!("/ipfs/{cid}").as_str())
.with_status(200)
.with_body(envelope)
.create_async()
.await;
let paths = recover_from_arweave(
&server.url(),
&server.url(),
"alice",
"myrepo",
&dest,
&reader,
)
.await
.unwrap();
assert_eq!(paths, vec!["secret/b.txt".to_string()]);
tx1.assert_async().await; // exactly one /TX1 fetch
}
/// A caller who is not a recipient cannot decrypt the envelope, so nothing is
/// recovered even though the manifest and envelope are reachable.
#[tokio::test]
+8
View File
@@ -19,6 +19,14 @@ services:
node:
build: .
# Reap reparented orphan processes so git's grandchildren (e.g. pack-objects
# left behind when a served clone disconnects) can't accumulate as zombies
# and exhaust the PID table (#53). The image also runs under tini, so this is
# belt-and-suspenders for the quickstart.
init: true
# Cap the process table so a regression fails loudly instead of taking the
# host down. Sits well above normal concurrent git fan-out.
pids_limit: 1024
depends_on:
postgres:
condition: service_healthy
+28
View File
@@ -141,6 +141,34 @@ During the cooldown your node still earns rewards if it keeps heartbeating.
---
## Hardening: owner-only push
By default the node authenticates every `git-receive-pack` push (a valid RFC 9421
`did:key` signature) but does **not** check that the pusher owns the repo, except
on branches that are explicitly protected. Because `did:key` is self-certifying,
any party can generate a key, derive its DID, sign, and push to an unprotected
branch — authentication is not authorization.
To require the authenticated pusher to be the repo owner on **every** branch, set:
```bash
GITLAWB_ENFORCE_OWNER_PUSH=true
```
- **Default `false`** — preserves current behavior so live nodes are unaffected by
an upgrade. Turn it on once you're ready for owner-only writes.
- **When `true`** — a push whose authenticated DID is not the repo owner is
rejected (HTTP 403) before any ref update is applied. The owner is matched in
both the full `did:key:z6Mk…` form and its bare `z6Mk…` suffix.
- **Caution: this blocks every non-owner pusher, including your own delegated and
CI agents.** Push authorization is owner-only today — a UCAN `git/push`
capability is verified but not yet honored for authorization, so delegated keys
cannot push while this is on. Don't enable it until every identity that pushes
to your repos is the owner, or you'll lock out your own automation. Scoped
collaborator / UCAN-delegated push rights are a planned follow-up.
---
## Operational checklist
| Concern | Recommendation |
+124
View File
@@ -0,0 +1,124 @@
<#
.SYNOPSIS
gitlawb installer for Windows - downloads pre-built binaries from GitHub Releases.
.EXAMPLE
irm https://gitlawb.com/install.ps1 | iex
.EXAMPLE
& ([scriptblock]::Create((irm https://gitlawb.com/install.ps1))) -Version v0.3.9
#>
[CmdletBinding()]
param(
[string]$Version = "latest"
)
$ErrorActionPreference = "Stop"
# Windows PowerShell 5.1 may default to TLS 1.0; GitHub requires TLS 1.2+.
[Net.ServicePointManager]::SecurityProtocol = [Net.ServicePointManager]::SecurityProtocol -bor [Net.SecurityProtocolType]::Tls12
$Repo = if ($env:GITLAWB_RELEASE_REPO) { $env:GITLAWB_RELEASE_REPO } else { "Gitlawb/node" }
$InstallDir = if ($env:GITLAWB_INSTALL_DIR) { $env:GITLAWB_INSTALL_DIR } else { "$env:LOCALAPPDATA\Programs\gitlawb" }
$arch = if ($env:PROCESSOR_ARCHITEW6432) {
$env:PROCESSOR_ARCHITEW6432
} else {
$env:PROCESSOR_ARCHITECTURE
}
if ($arch -ne "AMD64") {
throw "Unsupported architecture: $arch. Only x64 Windows binaries are published. Use WSL for arm64."
}
$target = "x86_64-pc-windows-msvc"
if ($Version -eq "latest") {
Write-Host "Fetching latest release version..."
$rel = Invoke-RestMethod -Uri "https://api.github.com/repos/$Repo/releases/latest" -UseBasicParsing -Headers @{ "User-Agent" = "gitlawb-installer" }
$tag = $rel.tag_name
} elseif ($Version.StartsWith("v")) {
$tag = $Version
} else {
$tag = "v$Version"
}
$ver = $tag.TrimStart("v")
$archive = "gitlawb-node-$ver-$target.zip"
$url = "https://github.com/$Repo/releases/download/$tag/$archive"
Write-Host "Installing gitlawb $tag for windows/x64"
Write-Host " Archive: $archive"
Write-Host " Into: $InstallDir"
$tmp = Join-Path ([System.IO.Path]::GetTempPath()) ("gitlawb-" + [System.Guid]::NewGuid().ToString("N"))
New-Item -ItemType Directory -Path $tmp | Out-Null
try {
$zipPath = Join-Path $tmp $archive
Write-Host "Downloading..."
Invoke-WebRequest -Uri $url -OutFile $zipPath -UseBasicParsing -Headers @{ "User-Agent" = "gitlawb-installer" }
# Verify checksum. Every published asset has a matching .sha256, so fail closed
# if it cannot be fetched rather than installing an unverified binary.
$sumPath = "$zipPath.sha256"
Invoke-WebRequest -Uri "$url.sha256" -OutFile $sumPath -UseBasicParsing -Headers @{ "User-Agent" = "gitlawb-installer" }
$expected = ((Get-Content $sumPath -Raw).Trim() -split '\s+')[0].ToLower()
$actual = (Get-FileHash $zipPath -Algorithm SHA256).Hash.ToLower()
if ($expected -ne $actual) {
throw "checksum mismatch! expected $expected got $actual"
}
Write-Host " checksum OK"
Write-Host "Extracting..."
$extract = Join-Path $tmp "extract"
Expand-Archive -Path $zipPath -DestinationPath $extract -Force
$pkgDir = Get-ChildItem -Path $extract -Directory | Select-Object -First 1
New-Item -ItemType Directory -Path $InstallDir -Force | Out-Null
# The Windows archive always ships the two CLI binaries; the daemon is optional.
$requiredBins = @("gl.exe", "git-remote-gitlawb.exe")
$optionalBins = @("gitlawb-node.exe")
$missing = $requiredBins | Where-Object {
-not (Test-Path (Join-Path $pkgDir.FullName $_))
}
if ($missing.Count -gt 0) {
throw "archive is missing required binaries: $($missing -join ', ')"
}
$installed = @()
foreach ($bin in $requiredBins + $optionalBins) {
$src = Join-Path $pkgDir.FullName $bin
if (Test-Path $src) {
Copy-Item -Path $src -Destination (Join-Path $InstallDir $bin) -Force
$installed += $bin
}
}
# Drop binaries no longer shipped (e.g. an optional one removed in a new release)
# so an upgrade in place never leaves a stale, mixed-version install behind.
foreach ($bin in $requiredBins + $optionalBins) {
if ($installed -notcontains $bin) {
$dst = Join-Path $InstallDir $bin
if (Test-Path $dst) { Remove-Item -Path $dst -Force }
}
}
Write-Host ""
Write-Host "Installed gitlawb $tag"
foreach ($bin in $installed) { Write-Host " $bin -> $InstallDir\$bin" }
}
finally {
Remove-Item -Recurse -Force $tmp -ErrorAction SilentlyContinue
}
# Add to the user PATH if missing.
$userPath = [Environment]::GetEnvironmentVariable("Path", "User")
if (($userPath -split ';') -notcontains $InstallDir) {
$newPath = if ([string]::IsNullOrEmpty($userPath)) { $InstallDir } else { "$userPath;$InstallDir" }
[Environment]::SetEnvironmentVariable("Path", $newPath, "User")
Write-Host ""
Write-Host "Added $InstallDir to your PATH. Restart your terminal, then run:"
} else {
Write-Host ""
Write-Host "Run:"
}
Write-Host " gl doctor"
Write-Host " gl quickstart"
Write-Host ""
Write-Host "Docs: https://docs.gitlawb.com"
+15
View File
@@ -0,0 +1,15 @@
node_modules/
# Platform binaries — laid in by the release workflow at publish time, not committed.
packages/gl-darwin-arm64/gl
packages/gl-darwin-arm64/git-remote-gitlawb
packages/gl-darwin-x64/gl
packages/gl-darwin-x64/git-remote-gitlawb
packages/gl-linux-arm64/gl
packages/gl-linux-arm64/git-remote-gitlawb
packages/gl-linux-x64/gl
packages/gl-linux-x64/git-remote-gitlawb
# Wrapper bin placeholders are overwritten by postinstall.
packages/gl/bin/gl
packages/gl/bin/git-remote-gitlawb
+37
View File
@@ -0,0 +1,37 @@
# npm distribution for the gitlawb CLI
This directory packages the `gl` and `git-remote-gitlawb` binaries as the
[`@gitlawb/gl`](https://www.npmjs.com/package/@gitlawb/gl) npm package.
It uses the **optionalDependencies platform-package** pattern (the same one
esbuild/swc/turbo use): the wrapper `@gitlawb/gl` declares four per-platform
packages as `optionalDependencies`, each gated by `os`/`cpu`. npm installs only
the one matching the host, and the wrapper's `postinstall` (`install.js`) copies
the binary out of it into `bin/`. No binaries are downloaded at install time.
```text
packages/
gl/ @gitlawb/gl wrapper (postinstall, bin shims)
gl-darwin-arm64/ @gitlawb/gl-darwin-arm64 aarch64-apple-darwin
gl-darwin-x64/ @gitlawb/gl-darwin-x64 x86_64-apple-darwin
gl-linux-arm64/ @gitlawb/gl-linux-arm64 aarch64-unknown-linux-musl
gl-linux-x64/ @gitlawb/gl-linux-x64 x86_64-unknown-linux-musl
```
## Publishing
Publishing is **automated** by `.github/workflows/release.yml` (the `npm-publish`
job) on every release-please release. That job:
1. Downloads the four unix release tarballs from the `Gitlawb/node` GitHub release.
2. Lays `gl` + `git-remote-gitlawb` into each platform package.
3. Rewrites every `package.json` `version` (and the wrapper's
`optionalDependencies` pins) to the release version.
4. Publishes the four platform packages first, then the wrapper, with
`npm publish --provenance --access public`.
The committed `version` fields here are placeholders; the workflow is the single
source of truth and always uses the release tag. There is no manual publish step.
Windows is intentionally not published to npm — the wrapper points Windows users
at the curl/PowerShell installer instead.
+21
View File
@@ -0,0 +1,21 @@
{
"name": "@gitlawb/gl-darwin-arm64",
"version": "0.3.9",
"description": "Gitlawb CLI binary for macOS Apple Silicon (arm64)",
"license": "MIT",
"os": ["darwin"],
"cpu": ["arm64"],
"publishConfig": {
"access": "public",
"registry": "https://registry.npmjs.org/"
},
"repository": {
"type": "git",
"url": "https://github.com/Gitlawb/node",
"directory": "npm/packages/gl-darwin-arm64"
},
"files": [
"gl",
"git-remote-gitlawb"
]
}
+21
View File
@@ -0,0 +1,21 @@
{
"name": "@gitlawb/gl-darwin-x64",
"version": "0.3.9",
"description": "Gitlawb CLI binary for macOS Intel (x64)",
"license": "MIT",
"os": ["darwin"],
"cpu": ["x64"],
"publishConfig": {
"access": "public",
"registry": "https://registry.npmjs.org/"
},
"repository": {
"type": "git",
"url": "https://github.com/Gitlawb/node",
"directory": "npm/packages/gl-darwin-x64"
},
"files": [
"gl",
"git-remote-gitlawb"
]
}
+21
View File
@@ -0,0 +1,21 @@
{
"name": "@gitlawb/gl-linux-arm64",
"version": "0.3.9",
"description": "Gitlawb CLI binary for Linux arm64",
"license": "MIT",
"os": ["linux"],
"cpu": ["arm64"],
"publishConfig": {
"access": "public",
"registry": "https://registry.npmjs.org/"
},
"repository": {
"type": "git",
"url": "https://github.com/Gitlawb/node",
"directory": "npm/packages/gl-linux-arm64"
},
"files": [
"gl",
"git-remote-gitlawb"
]
}
+21
View File
@@ -0,0 +1,21 @@
{
"name": "@gitlawb/gl-linux-x64",
"version": "0.3.9",
"description": "Gitlawb CLI binary for Linux x64",
"license": "MIT",
"os": ["linux"],
"cpu": ["x64"],
"publishConfig": {
"access": "public",
"registry": "https://registry.npmjs.org/"
},
"repository": {
"type": "git",
"url": "https://github.com/Gitlawb/node",
"directory": "npm/packages/gl-linux-x64"
},
"files": [
"gl",
"git-remote-gitlawb"
]
}
+67
View File
@@ -0,0 +1,67 @@
# @gitlawb/gl
The [gitlawb](https://gitlawb.com) CLI — decentralized git for AI agents and developers.
## Install
```bash
# macOS / Linux / WSL
npm install -g @gitlawb/gl
```
Also works with yarn, pnpm, and bun:
```bash
# macOS / Linux / WSL
yarn global add @gitlawb/gl
pnpm add -g @gitlawb/gl
bun add -g @gitlawb/gl
```
Native Windows is not supported via npm — use the PowerShell installer below.
### Other install methods
```bash
# Homebrew
brew install gitlawb/tap/gl
# curl (macOS / Linux)
curl -sSf https://gitlawb.com/install.sh | sh
# PowerShell (Windows)
irm https://gitlawb.com/install.ps1 | iex
```
## Quick start
```bash
gl identity new
gl register
gl doctor
gl repo create my-project --description "My first gitlawb repo"
git remote add gitlawb gitlawb://my-project
git push gitlawb main
```
## What's included
- **`gl`** — the main CLI for identity, repos, PRs, bounties, and agents
- **`git-remote-gitlawb`** — git remote helper for `gitlawb://` URLs
## Supported platforms
| Platform | Architecture | Package |
|----------|-------------|---------|
| macOS | Apple Silicon (arm64) | `@gitlawb/gl-darwin-arm64` |
| macOS | Intel (x64) | `@gitlawb/gl-darwin-x64` |
| Linux | arm64 | `@gitlawb/gl-linux-arm64` |
| Linux | x64 | `@gitlawb/gl-linux-x64` |
Windows is supported via the PowerShell installer or WSL.
## Links
- Docs: https://docs.gitlawb.com
- Site: https://gitlawb.com
- Source: https://github.com/Gitlawb/node
View File
Whitespace-only changes.
+89
View File
@@ -0,0 +1,89 @@
#!/usr/bin/env node
const { existsSync, copyFileSync, chmodSync, mkdirSync } = require("fs");
const { join, dirname } = require("path");
const PLATFORM_PACKAGES = {
"darwin-arm64": "@gitlawb/gl-darwin-arm64",
"darwin-x64": "@gitlawb/gl-darwin-x64",
"linux-arm64": "@gitlawb/gl-linux-arm64",
"linux-x64": "@gitlawb/gl-linux-x64",
};
const BINARIES = ["gl", "git-remote-gitlawb"];
function getPlatformKey() {
const platform = process.platform;
const arch = process.arch;
if (arch !== "arm64" && arch !== "x64") {
throw new Error(
`Unsupported architecture: ${arch}. Gitlawb CLI supports x64 and arm64.`
);
}
if (platform !== "darwin" && platform !== "linux") {
throw new Error(
`Unsupported platform: ${platform}. Gitlawb CLI supports macOS and Linux.\n` +
`For Windows, use WSL or the installer: https://gitlawb.com/install.ps1`
);
}
return `${platform}-${arch}`;
}
function findPlatformPackage(packageName) {
try {
const pkgJson = require.resolve(`${packageName}/package.json`);
return dirname(pkgJson);
} catch {
return null;
}
}
function install() {
const platformKey = getPlatformKey();
const packageName = PLATFORM_PACKAGES[platformKey];
if (!packageName) {
console.error(`No binary package available for ${platformKey}`);
process.exit(1);
}
const packageDir = findPlatformPackage(packageName);
if (!packageDir) {
// On a supported platform the matching optional dependency must be present;
// fail loudly rather than leave @gitlawb/gl installed without its binaries.
console.error(
`@gitlawb/gl: Platform package ${packageName} not found.\n` +
`This can happen if optional dependencies were skipped.\n` +
`Install manually: curl -sSf https://gitlawb.com/install.sh | sh`
);
process.exit(1);
}
const binDir = join(__dirname, "bin");
mkdirSync(binDir, { recursive: true });
for (const binary of BINARIES) {
const src = join(packageDir, binary);
const dest = join(binDir, binary);
if (!existsSync(src)) {
throw new Error(`Binary ${binary} not found in ${packageName}`);
}
copyFileSync(src, dest);
chmodSync(dest, 0o755);
}
console.log(`@gitlawb/gl: Installed gitlawb CLI for ${platformKey}`);
}
try {
install();
} catch (err) {
console.error(`@gitlawb/gl: ${err.message}`);
process.exit(1);
}
+42
View File
@@ -0,0 +1,42 @@
{
"name": "@gitlawb/gl",
"version": "0.3.9",
"description": "Gitlawb CLI — decentralized git for AI agents and developers",
"license": "MIT",
"bin": {
"gl": "bin/gl",
"git-remote-gitlawb": "bin/git-remote-gitlawb"
},
"scripts": {
"postinstall": "node install.js"
},
"publishConfig": {
"access": "public",
"registry": "https://registry.npmjs.org/"
},
"repository": {
"type": "git",
"url": "https://github.com/Gitlawb/node",
"directory": "npm/packages/gl"
},
"keywords": [
"gitlawb",
"git",
"decentralized",
"cli",
"ai-agents",
"did",
"ucan"
],
"files": [
"bin",
"install.js",
"README.md"
],
"optionalDependencies": {
"@gitlawb/gl-darwin-arm64": "0.3.9",
"@gitlawb/gl-darwin-x64": "0.3.9",
"@gitlawb/gl-linux-arm64": "0.3.9",
"@gitlawb/gl-linux-x64": "0.3.9"
}
}
+9 -2
View File
@@ -2,11 +2,18 @@
"$schema": "https://raw.githubusercontent.com/googleapis/release-please/main/schemas/config.json",
"packages": {
".": {
"release-type": "rust",
"release-type": "simple",
"package-name": "gitlawb-node",
"bump-minor-pre-major": true,
"include-v-in-tag": true,
"include-component-in-tag": false
"include-component-in-tag": false,
"extra-files": [
{ "type": "generic", "path": "crates/gitlawb-core/Cargo.toml" },
{ "type": "generic", "path": "crates/gitlawb-node/Cargo.toml" },
{ "type": "generic", "path": "crates/gl/Cargo.toml" },
{ "type": "generic", "path": "crates/git-remote-gitlawb/Cargo.toml" },
{ "type": "generic", "path": "crates/gitlawb-attest/Cargo.toml" }
]
}
}
}